def test_editlinksnode(self):
from ...registry.tests import small_testing_registry
reg = small_testing_registry()
file_desc = reg.widget("Orange.widgets.data.owfile.OWFile")
bayes_desc = reg.widget("Orange.widgets.classify.ownaivebayes."
"OWNaiveBayes")
source_node = SchemeNode(file_desc, title="This is File")
sink_node = SchemeNode(bayes_desc)
scene = QGraphicsScene()
view = QGraphicsView(scene)
node = EditLinksNode(node=source_node)
scene.addItem(node)
node = EditLinksNode(direction=Qt.RightToLeft)
node.setSchemeNode(sink_node)
node.setPos(300, 0)
scene.addItem(node)
view.show()
view.resize(800, 300)
self.app.exec_()
def test_other_exporter(self):
sc = QGraphicsScene()
sc.addItem(QGraphicsRectItem(0, 0, 3, 3))
with patch("orangewidget.io.ImgFormat._get_exporter", Mock()) as mfn:
with self.assertRaises(Exception):
imgio.ImgFormat.write("", sc)
self.assertEqual(0, mfn.call_count)
def test_editlinksnode(self):
from ...registry.tests import small_testing_registry
reg = small_testing_registry()
file_desc = reg.widget("Orange.widgets.data.owfile.OWFile")
bayes_desc = reg.widget("Orange.widgets.classify.ownaivebayes."
"OWNaiveBayes")
source_node = SchemeNode(file_desc, title="This is File")
sink_node = SchemeNode(bayes_desc)
scene = QGraphicsScene()
view = QGraphicsView(scene)
node = EditLinksNode(node=source_node)
scene.addItem(node)
node = EditLinksNode(direction=Qt.RightToLeft)
node.setSchemeNode(sink_node)
node.setPos(300, 0)
scene.addItem(node)
view.show()
view.resize(800, 300)
self.app.exec_()
class IconWidget(QWidget):
def __init__(self):
super().__init__()
self.setLayout(QHBoxLayout())
self.layout().setContentsMargins(0, 0, 0, 0)
self.setFixedSize(50, 50)
view = QGraphicsView()
self.layout().addWidget(view)
self.scene = QGraphicsScene(view)
view.setScene(self.scene)
def set_widget(self, widget_description, category_description):
node = NodeItem(widget_description)
if category_description is not None:
node.setWidgetCategory(category_description)
self.scene.addItem(node)
def render_as_png(self, filename):
png = self.__transparent_png()
painter = QtGui.QPainter(png)
painter.setRenderHint(QtGui.QPainter.Antialiasing, 1)
self.scene.render(painter, QRectF(0, 0, 50, 50),
QRectF(-25, -25, 50, 50))
painter.end()
png.save(filename)
def __transparent_png(self):
# PyQt is stupid and does not increment reference count on bg
self.__bg = bg = np.zeros((50, 50, 4), dtype=np.ubyte)
return QImage(bg.ctypes.data, bg.shape[1], bg.shape[0],
QtGui.QImage.Format_ARGB32)
class TestGraphicsPixmapWidget(GuiTest):
def setUp(self) -> None:
super().setUp()
self.scene = QGraphicsScene()
self.view = QGraphicsView(self.scene)
def tearDown(self) -> None:
self.scene.clear()
self.scene.deleteLater()
self.view.deleteLater()
del self.scene
del self.view
def test_graphicspixmapwidget(self):
w = GraphicsPixmapWidget()
self.scene.addItem(w)
w.setPixmap(QPixmap(100, 100))
p = w.pixmap()
self.assertEqual(p.size(), QSize(100, 100))
self.view.grab()
w.setScaleContents(True)
w.setAspectRatioMode(Qt.KeepAspectRatio)
s = w.sizeHint(Qt.PreferredSize)
self.assertEqual(s, QSizeF(100., 100.))
s = w.sizeHint(Qt.PreferredSize, QSizeF(200., -1.))
self.assertEqual(s, QSizeF(200., 200.))
s = w.sizeHint(Qt.PreferredSize, QSizeF(-1., 200.))
self.assertEqual(s, QSizeF(200., 200.))
self.view.grab()
class IconWidget(QWidget):
def __init__(self):
super().__init__()
self.setLayout(QHBoxLayout())
self.layout().setContentsMargins(0, 0, 0, 0)
self.setFixedSize(50, 50)
view = QGraphicsView()
self.layout().addWidget(view)
self.scene = QGraphicsScene(view)
view.setScene(self.scene)
def set_widget(self, widget_description, category_description):
node = NodeItem(widget_description)
if category_description is not None:
node.setWidgetCategory(category_description)
self.scene.addItem(node)
def render_as_png(self, filename):
img = QImage(50, 50, QImage.Format_ARGB32)
img.fill(Qt.transparent)
painter = QPainter(img)
painter.setRenderHint(QPainter.Antialiasing, 1)
self.scene.render(painter, QRectF(0, 0, 50, 50),
QRectF(-25, -25, 50, 50))
painter.end()
img.save(filename)
def paint(self, painter, option, index):
# type: (QPainter, QStyleOptionViewItem, QModelIndex) -> None
data = index.data(FeatureStatisticsTableModel.DistributionRole)
if data is None:
return super().paint(painter, option, index)
row = index.model().mapToSourceRows(index.row())
if row not in self.__cache:
scene = QGraphicsScene(self)
attribute, data = data
histogram = Histogram(
data=data,
variable=attribute,
color_attribute=self.color_attribute,
border=(0, 0, 2, 0),
border_color='#ccc',
)
scene.addItem(histogram)
self.__cache[row] = scene
painter.setRenderHint(QPainter.HighQualityAntialiasing)
if option.state & QStyle.State_Selected:
background_color = option.palette.highlight()
else:
background_color = index.data(Qt.BackgroundRole)
self.__cache[row].setBackgroundBrush(background_color)
self.__cache[row].render(
painter,
target=QRectF(option.rect),
mode=Qt.IgnoreAspectRatio,
)
def main(argv=None): # pragma: no cover
# pylint: disable=import-outside-toplevel
import sys
from AnyQt.QtWidgets import QGraphicsScene, QMenu
from AnyQt.QtGui import QBrush
app = QApplication(argv or sys.argv)
scene = QGraphicsScene()
view = GraphicsWidgetView(scene)
scene.setParent(view)
view.setContextMenuPolicy(Qt.CustomContextMenu)
def context(pos):
menu = QMenu(view)
menu.addActions(view.actions())
a = menu.addAction("Aspect mode")
am = QMenu(menu)
am.addAction("Ignore", lambda: view.setAspectMode(Qt.IgnoreAspectRatio))
am.addAction("Keep", lambda: view.setAspectMode(Qt.KeepAspectRatio))
am.addAction("Keep by expanding", lambda: view.setAspectMode(Qt.KeepAspectRatioByExpanding))
a.setMenu(am)
menu.popup(view.viewport().mapToGlobal(pos))
view.customContextMenuRequested.connect(context)
w = QGraphicsWidget()
w.setPreferredSize(500, 500)
palette = w.palette()
palette.setBrush(palette.Window, QBrush(Qt.red, Qt.BDiagPattern))
w.setPalette(palette)
w.setAutoFillBackground(True)
scene.addItem(w)
view.setCentralWidget(w)
view.show()
return app.exec()
def data(self, index, role=Qt.DisplayRole):
# type: (QModelIndex, Qt.QDisplayRole) -> Any
if not index.isValid():
return None
idx = index.row()
if role == Qt.SizeHintRole:
return self.item_scale * QSize(100, 100)
if role == Qt.DisplayRole:
if 'tree' not in self._other_data[idx]:
scene = QGraphicsScene(parent=self)
tree = PythagorasTreeViewer(
adapter=self._list[idx],
weight_adjustment=OWPythagoreanForest.SIZE_CALCULATION[
self.size_calc_idx][1],
interactive=False,
padding=100,
depth_limit=self.depth_limit,
target_class_index=self.target_class_idx,
)
scene.addItem(tree)
self._other_data[idx]['scene'] = scene
self._other_data[idx]['tree'] = tree
return self._other_data[idx]['scene']
return super().data(index, role)
def data(self, index, role=Qt.DisplayRole):
# type: (QModelIndex, Qt.QDisplayRole) -> Any
if not index.isValid():
return None
idx = index.row()
if role == Qt.SizeHintRole:
return self.item_scale * QSize(100, 100)
if role == Qt.DisplayRole:
if 'tree' not in self._other_data[idx]:
scene = QGraphicsScene(parent=self)
tree = PythagorasTreeViewer(
adapter=self._list[idx],
weight_adjustment=OWPythagoreanForest.SIZE_CALCULATION[
self.size_calc_idx][1],
interactive=False,
padding=100,
depth_limit=self.depth_limit,
target_class_index=self.target_class_idx,
)
scene.addItem(tree)
self._other_data[idx]['scene'] = scene
self._other_data[idx]['tree'] = tree
return self._other_data[idx]['scene']
return super().data(index, role)
def display():
# pylint: disable=too-many-branches
def format_zeros(str_val):
"""Zeros should be handled separately as they cannot be negative."""
if float(str_val) == 0:
num_decimals = min(self.variables[row].number_of_decimals,
2)
str_val = f"{0:.{num_decimals}f}"
return str_val
def render_value(value):
if np.isnan(value):
return ""
if np.isinf(value):
return "∞"
str_val = attribute.str_val(value)
if attribute.is_continuous and not attribute.is_time:
str_val = format_zeros(str_val)
return str_val
if column == self.Columns.NAME:
return attribute.name
elif column == self.Columns.DISTRIBUTION:
if isinstance(attribute,
(DiscreteVariable, ContinuousVariable)):
if row not in self.__distributions_cache:
scene = QGraphicsScene(parent=self)
histogram = Histogram(
data=self.table,
variable=attribute,
color_attribute=self.target_var,
border=(0, 0, 2, 0),
border_color='#ccc',
)
scene.addItem(histogram)
self.__distributions_cache[row] = scene
return self.__distributions_cache[row]
elif column == self.Columns.CENTER:
return render_value(self._center[row])
elif column == self.Columns.MEDIAN:
return render_value(self._median[row])
elif column == self.Columns.DISPERSION:
if isinstance(attribute, TimeVariable):
return format_time_diff(self._min[row], self._max[row])
elif isinstance(attribute, DiscreteVariable):
return "%.3g" % self._dispersion[row]
else:
return render_value(self._dispersion[row])
elif column == self.Columns.MIN:
if not isinstance(attribute, DiscreteVariable):
return render_value(self._min[row])
elif column == self.Columns.MAX:
if not isinstance(attribute, DiscreteVariable):
return render_value(self._max[row])
elif column == self.Columns.MISSING:
return '%d (%d%%)' % (self._missing[row], 100 *
self._missing[row] / self.n_instances)
return None
def test_other_exporter(self):
sc = QGraphicsScene()
sc.addItem(QGraphicsRectItem(0, 0, 3, 3))
with patch("Orange.widgets.io.ImgFormat._get_exporter",
Mock()) as mfn:
with self.assertRaises(Exception):
imgio.ImgFormat.write("", sc)
self.assertEqual(0, mfn.call_count)
def test_pdf(self):
sc = QGraphicsScene()
sc.addItem(QGraphicsRectItem(0, 0, 20, 20))
fd, fname = tempfile.mkstemp()
os.close(fd)
try:
imgio.PdfFormat.write_image(fname, sc)
finally:
os.unlink(fname)
def test_other(self):
fd, fname = tempfile.mkstemp('.pdf')
os.close(fd)
sc = QGraphicsScene()
sc.addItem(QGraphicsRectItem(0, 0, 3, 3))
try:
imgio.PdfFormat.write(fname, sc)
with open(fname, "rb") as f:
self.assertTrue(f.read().startswith(b'%PDF'))
finally:
os.unlink(fname)
def test_graphicstextwidget(self):
scene = QGraphicsScene()
view = QGraphicsView(scene)
text = GraphicsTextWidget()
text.setHtml("<center><b>a text</b></center><p>paragraph</p>")
scene.addItem(text)
view.show()
view.resize(400, 300)
self.app.exec_()
def test_other(self):
fd, fname = tempfile.mkstemp('.png')
os.close(fd)
sc = QGraphicsScene()
sc.addItem(QGraphicsRectItem(0, 0, 3, 3))
try:
imgio.PngFormat.write(fname, sc)
im = QImage(fname)
# writer adds 15*2 of empty space
self.assertEqual((30+4, 30+4), (im.width(), im.height()))
finally:
os.unlink(fname)
class TestDendrogramWidget(GuiTest):
def setUp(self) -> None:
super().setUp()
self.scene = QGraphicsScene()
self.view = QGraphicsView(self.scene)
self.widget = DendrogramWidget()
self.scene.addItem(self.widget)
def tearDown(self) -> None:
self.scene.clear()
del self.widget
del self.view
super().tearDown()
def test_widget(self):
w = self.widget
T = hierarchical.Tree
C = hierarchical.ClusterData
S = hierarchical.SingletonData
def t(h: float, left: T, right: T):
return T(C((left.value.first, right.value.last), h), (left, right))
def leaf(r, index):
return T(S((r, r + 1), 0.0, index))
T = hierarchical.Tree
w.set_root(t(0.0, leaf(0, 0), leaf(1, 1)))
w.resize(w.effectiveSizeHint(Qt.PreferredSize))
h = w.height_at(QPoint())
self.assertEqual(h, 0)
h = w.height_at(QPoint(10, 0))
self.assertEqual(h, 0)
self.assertEqual(w.pos_at_height(0).x(), w.rect().x())
self.assertEqual(w.pos_at_height(1).x(), w.rect().x())
height = np.finfo(float).eps
w.set_root(t(height, leaf(0, 0), leaf(1, 1)))
h = w.height_at(QPoint())
self.assertEqual(h, height)
h = w.height_at(QPoint(w.size().width(), 0))
self.assertEqual(h, 0)
self.assertEqual(w.pos_at_height(0).x(), w.rect().right())
self.assertEqual(w.pos_at_height(height).x(), w.rect().left())
def __init__(self, master, *args):
QGraphicsView.__init__(self, *args)
self.master = master
self.setHorizontalScrollBarPolicy(Qt.ScrollBarAsNeeded)
self.setVerticalScrollBarPolicy(Qt.ScrollBarAsNeeded)
self.setRenderHints(QPainter.Antialiasing)
scene = QGraphicsScene(self)
self.pixmapGraphicsItem = QGraphicsPixmapItem(None)
scene.addItem(self.pixmapGraphicsItem)
self.setScene(scene)
self.setMouseTracking(True)
self.viewport().setMouseTracking(True)
self.setFocusPolicy(Qt.WheelFocus)
def __init__(self, master, *args):
QGraphicsView.__init__(self, *args)
self.master = master
self.setHorizontalScrollBarPolicy(Qt.ScrollBarAsNeeded)
self.setVerticalScrollBarPolicy(Qt.ScrollBarAsNeeded)
self.setRenderHints(QPainter.Antialiasing)
scene = QGraphicsScene(self)
self.pixmapGraphicsItem = QGraphicsPixmapItem(None)
scene.addItem(self.pixmapGraphicsItem)
self.setScene(scene)
self.setMouseTracking(True)
self.viewport().setMouseTracking(True)
self.setFocusPolicy(Qt.WheelFocus)
def display():
# pylint: disable=too-many-branches
def render_value(value):
return "" if np.isnan(value) else attribute.str_val(value)
if column == self.Columns.NAME:
return attribute.name
elif column == self.Columns.DISTRIBUTION:
if isinstance(attribute,
(DiscreteVariable, ContinuousVariable)):
if row not in self.__distributions_cache:
scene = QGraphicsScene(parent=self)
histogram = Histogram(
data=self.table,
variable=attribute,
color_attribute=self.target_var,
border=(0, 0, 2, 0),
border_color='#ccc',
)
scene.addItem(histogram)
self.__distributions_cache[row] = scene
return self.__distributions_cache[row]
elif column == self.Columns.CENTER:
return render_value(self._center[row])
elif column == self.Columns.DISPERSION:
if isinstance(attribute, TimeVariable):
return format_time_diff(self._min[row], self._max[row])
elif isinstance(attribute, DiscreteVariable):
return "%.3g" % self._dispersion[row]
else:
return render_value(self._dispersion[row])
elif column == self.Columns.MIN:
if not isinstance(attribute, DiscreteVariable) \
or attribute.ordered:
return render_value(self._min[row])
elif column == self.Columns.MAX:
if not isinstance(attribute, DiscreteVariable) \
or attribute.ordered:
return render_value(self._max[row])
elif column == self.Columns.MISSING:
return '%d (%d%%)' % (
self._missing[row],
100 * self._missing[row] / self.n_instances
)
return None
def render_drop_shadow_frame(pixmap, shadow_rect, shadow_color, offset, radius,
rect_fill_color):
pixmap.fill(QColor(0, 0, 0, 0))
scene = QGraphicsScene()
rect = QGraphicsRectItem(shadow_rect)
rect.setBrush(QColor(rect_fill_color))
rect.setPen(QPen(Qt.NoPen))
scene.addItem(rect)
effect = QGraphicsDropShadowEffect(color=shadow_color,
blurRadius=radius,
offset=offset)
rect.setGraphicsEffect(effect)
scene.setSceneRect(QRectF(QPointF(0, 0), QSizeF(pixmap.size())))
painter = QPainter(pixmap)
scene.render(painter)
painter.end()
scene.clear()
scene.deleteLater()
return pixmap
def render_drop_shadow_frame(pixmap, shadow_rect, shadow_color,
offset, radius, rect_fill_color):
pixmap.fill(QColor(0, 0, 0, 0))
scene = QGraphicsScene()
rect = QGraphicsRectItem(shadow_rect)
rect.setBrush(QColor(rect_fill_color))
rect.setPen(QPen(Qt.NoPen))
scene.addItem(rect)
effect = QGraphicsDropShadowEffect(color=shadow_color,
blurRadius=radius,
offset=offset)
rect.setGraphicsEffect(effect)
scene.setSceneRect(QRectF(QPointF(0, 0), QSizeF(pixmap.size())))
painter = QPainter(pixmap)
scene.render(painter)
painter.end()
scene.clear()
scene.deleteLater()
return pixmap
def save_widget_icon(
background,
icon: str,
outname: str,
export_size=QSize(100, 100),
format="png",
):
# create fake windget and category descriptions
widget_description = WidgetDescription("",
"",
icon=icon,
qualified_name="orangecanvas")
category = CategoryDescription(background=background)
item = NodeItem(widget_description)
item.setWidgetCategory(category)
iconItem = item.icon_item
shapeItem = item.shapeItem
shapeSize = export_size
iconSize = QSize(export_size.width() * 3 / 4, export_size.height() * 3 / 4)
rect = QRectF(QPointF(-shapeSize.width() / 2, -shapeSize.height() / 2),
QSizeF(shapeSize))
shapeItem.setShapeRect(rect)
iconItem.setIconSize(iconSize)
iconItem.setPos(-iconSize.width() / 2, -iconSize.height() / 2)
image = QImage(export_size, QImage.Format_ARGB32)
image.fill(QColor("#00FFFFFF"))
painter = QPainter(image)
painter.setRenderHint(QPainter.Antialiasing, 1)
scene = QGraphicsScene()
scene.addItem(shapeItem)
scene.render(painter, QRectF(image.rect()), scene.sceneRect())
painter.end()
if not image.save(outname, format, 80):
print("Failed to save " + outname)
def test_editlinksnode(self):
reg = small_testing_registry()
one_desc = reg.widget("one")
negate_desc = reg.widget("negate")
source_node = SchemeNode(one_desc, title="This is 1")
sink_node = SchemeNode(negate_desc)
scene = QGraphicsScene()
view = QGraphicsView(scene)
node = EditLinksNode(node=source_node)
scene.addItem(node)
node = EditLinksNode(direction=Qt.RightToLeft)
node.setSchemeNode(sink_node)
node.setPos(300, 0)
scene.addItem(node)
view.show()
view.resize(800, 300)
self.app.exec_()
class OWSilhouettePlot(widget.OWWidget):
name = "Silhouette Plot"
description = "Visually assess cluster quality and " \
"the degree of cluster membership."
icon = "icons/SilhouettePlot.svg"
priority = 300
keywords = []
class Inputs:
data = Input("Data", Orange.data.Table)
class Outputs:
selected_data = Output("Selected Data", Orange.data.Table, default=True)
annotated_data = Output(ANNOTATED_DATA_SIGNAL_NAME, Orange.data.Table)
replaces = [
"orangecontrib.prototypes.widgets.owsilhouetteplot.OWSilhouettePlot",
"Orange.widgets.unsupervised.owsilhouetteplot.OWSilhouettePlot"
]
settingsHandler = settings.PerfectDomainContextHandler()
#: Distance metric index
distance_idx = settings.Setting(0)
#: Group/cluster variable index
cluster_var_idx = settings.ContextSetting(0)
#: Annotation variable index
annotation_var_idx = settings.ContextSetting(0)
#: Group the (displayed) silhouettes by cluster
group_by_cluster = settings.Setting(True)
#: A fixed size for an instance bar
bar_size = settings.Setting(3)
#: Add silhouette scores to output data
add_scores = settings.Setting(False)
auto_commit = settings.Setting(True)
Distances = [("Euclidean", Orange.distance.Euclidean),
("Manhattan", Orange.distance.Manhattan),
("Cosine", Orange.distance.Cosine)]
graph_name = "scene"
buttons_area_orientation = Qt.Vertical
class Error(widget.OWWidget.Error):
need_two_clusters = Msg("Need at least two non-empty clusters")
singleton_clusters_all = Msg("All clusters are singletons")
memory_error = Msg("Not enough memory")
value_error = Msg("Distances could not be computed: '{}'")
class Warning(widget.OWWidget.Warning):
missing_cluster_assignment = Msg(
"{} instance{s} omitted (missing cluster assignment)")
nan_distances = Msg("{} instance{s} omitted (undefined distances)")
ignoring_categorical = Msg("Ignoring categorical features")
def __init__(self):
super().__init__()
#: The input data
self.data = None # type: Optional[Orange.data.Table]
#: Distance matrix computed from data
self._matrix = None # type: Optional[Orange.misc.DistMatrix]
#: An bool mask (size == len(data)) indicating missing group/cluster
#: assignments
self._mask = None # type: Optional[np.ndarray]
#: An array of cluster/group labels for instances with valid group
#: assignment
self._labels = None # type: Optional[np.ndarray]
#: An array of silhouette scores for instances with valid group
#: assignment
self._silhouette = None # type: Optional[np.ndarray]
self._silplot = None # type: Optional[SilhouettePlot]
gui.comboBox(
self.controlArea, self, "distance_idx", box="Distance",
items=[name for name, _ in OWSilhouettePlot.Distances],
orientation=Qt.Horizontal, callback=self._invalidate_distances)
box = gui.vBox(self.controlArea, "Cluster Label")
self.cluster_var_cb = gui.comboBox(
box, self, "cluster_var_idx", contentsLength=14, addSpace=4,
callback=self._invalidate_scores
)
gui.checkBox(
box, self, "group_by_cluster", "Group by cluster",
callback=self._replot)
self.cluster_var_model = itemmodels.VariableListModel(parent=self)
self.cluster_var_cb.setModel(self.cluster_var_model)
box = gui.vBox(self.controlArea, "Bars")
gui.widgetLabel(box, "Bar width:")
gui.hSlider(
box, self, "bar_size", minValue=1, maxValue=10, step=1,
callback=self._update_bar_size, addSpace=6)
gui.widgetLabel(box, "Annotations:")
self.annotation_cb = gui.comboBox(
box, self, "annotation_var_idx", contentsLength=14,
callback=self._update_annotations)
self.annotation_var_model = itemmodels.VariableListModel(parent=self)
self.annotation_var_model[:] = ["None"]
self.annotation_cb.setModel(self.annotation_var_model)
ibox = gui.indentedBox(box, 5)
self.ann_hidden_warning = warning = gui.widgetLabel(
ibox, "(increase the width to show)")
ibox.setFixedWidth(ibox.sizeHint().width())
warning.setVisible(False)
gui.rubber(self.controlArea)
gui.separator(self.buttonsArea)
box = gui.vBox(self.buttonsArea, "Output")
# Thunk the call to commit to call conditional commit
gui.checkBox(box, self, "add_scores", "Add silhouette scores",
callback=lambda: self.commit())
gui.auto_commit(
box, self, "auto_commit", "Commit",
auto_label="Auto commit", box=False)
# Ensure that the controlArea is not narrower than buttonsArea
self.controlArea.layout().addWidget(self.buttonsArea)
self.scene = QGraphicsScene()
self.view = QGraphicsView(self.scene)
self.view.setRenderHint(QPainter.Antialiasing, True)
self.view.setAlignment(Qt.AlignTop | Qt.AlignLeft)
self.mainArea.layout().addWidget(self.view)
def sizeHint(self):
sh = self.controlArea.sizeHint()
return sh.expandedTo(QSize(600, 720))
@Inputs.data
@check_sql_input
def set_data(self, data):
"""
Set the input dataset.
"""
self.closeContext()
self.clear()
error_msg = ""
warning_msg = ""
candidatevars = []
if data is not None:
candidatevars = [
v for v in data.domain.variables + data.domain.metas
if v.is_discrete and len(v.values) >= 2]
if not candidatevars:
error_msg = "Input does not have any suitable labels."
data = None
self.data = data
if data is not None:
self.cluster_var_model[:] = candidatevars
if data.domain.class_var in candidatevars:
self.cluster_var_idx = \
candidatevars.index(data.domain.class_var)
else:
self.cluster_var_idx = 0
annotvars = [var for var in data.domain.metas if var.is_string]
self.annotation_var_model[:] = ["None"] + annotvars
self.annotation_var_idx = 1 if len(annotvars) else 0
self.openContext(Orange.data.Domain(candidatevars))
self.error(error_msg)
self.warning(warning_msg)
def handleNewSignals(self):
if self.data is not None:
self._update()
self._replot()
self.unconditional_commit()
def clear(self):
"""
Clear the widget state.
"""
self.data = None
self._matrix = None
self._mask = None
self._silhouette = None
self._labels = None
self.cluster_var_model[:] = []
self.annotation_var_model[:] = ["None"]
self._clear_scene()
self.Error.clear()
self.Warning.clear()
def _clear_scene(self):
# Clear the graphics scene and associated objects
self.scene.clear()
self.scene.setSceneRect(QRectF())
self._silplot = None
def _invalidate_distances(self):
# Invalidate the computed distance matrix and recompute the silhouette.
self._matrix = None
self._invalidate_scores()
def _invalidate_scores(self):
# Invalidate and recompute the current silhouette scores.
self._labels = self._silhouette = self._mask = None
self._update()
self._replot()
if self.data is not None:
self.commit()
def _update(self):
# Update/recompute the distances/scores as required
self._clear_messages()
if self.data is None or not len(self.data):
self._reset_all()
return
if self._matrix is None and self.data is not None:
_, metric = self.Distances[self.distance_idx]
data = self.data
if not metric.supports_discrete and any(
a.is_discrete for a in data.domain.attributes):
self.Warning.ignoring_categorical()
data = Orange.distance.remove_discrete_features(data)
try:
self._matrix = np.asarray(metric(data))
except MemoryError:
self.Error.memory_error()
return
except ValueError as err:
self.Error.value_error(str(err))
return
self._update_labels()
def _reset_all(self):
self._mask = None
self._silhouette = None
self._labels = None
self._matrix = None
self._clear_scene()
def _clear_messages(self):
self.Error.clear()
self.Warning.clear()
def _update_labels(self):
labelvar = self.cluster_var_model[self.cluster_var_idx]
labels, _ = self.data.get_column_view(labelvar)
labels = np.asarray(labels, dtype=float)
cluster_mask = np.isnan(labels)
dist_mask = np.isnan(self._matrix).all(axis=0)
mask = cluster_mask | dist_mask
labels = labels.astype(int)
labels = labels[~mask]
labels_unq, _ = np.unique(labels, return_counts=True)
if len(labels_unq) < 2:
self.Error.need_two_clusters()
labels = silhouette = mask = None
elif len(labels_unq) == len(labels):
self.Error.singleton_clusters_all()
labels = silhouette = mask = None
else:
silhouette = sklearn.metrics.silhouette_samples(
self._matrix[~mask, :][:, ~mask], labels, metric="precomputed")
self._mask = mask
self._labels = labels
self._silhouette = silhouette
if mask is not None:
count_missing = np.count_nonzero(cluster_mask)
if count_missing:
self.Warning.missing_cluster_assignment(
count_missing, s="s" if count_missing > 1 else "")
count_nandist = np.count_nonzero(dist_mask)
if count_nandist:
self.Warning.nan_distances(
count_nandist, s="s" if count_nandist > 1 else "")
def _set_bar_height(self):
visible = self.bar_size >= 5
self._silplot.setBarHeight(self.bar_size)
self._silplot.setRowNamesVisible(visible)
self.ann_hidden_warning.setVisible(
not visible and self.annotation_var_idx > 0)
def _replot(self):
# Clear and replot/initialize the scene
self._clear_scene()
if self._silhouette is not None and self._labels is not None:
var = self.cluster_var_model[self.cluster_var_idx]
self._silplot = silplot = SilhouettePlot()
self._set_bar_height()
if self.group_by_cluster:
silplot.setScores(self._silhouette, self._labels, var.values,
var.colors)
else:
silplot.setScores(
self._silhouette,
np.zeros(len(self._silhouette), dtype=int),
[""], np.array([[63, 207, 207]])
)
self.scene.addItem(silplot)
self._update_annotations()
silplot.selectionChanged.connect(self.commit)
silplot.layout().activate()
self._update_scene_rect()
silplot.geometryChanged.connect(self._update_scene_rect)
def _update_bar_size(self):
if self._silplot is not None:
self._set_bar_height()
def _update_annotations(self):
if 0 < self.annotation_var_idx < len(self.annotation_var_model):
annot_var = self.annotation_var_model[self.annotation_var_idx]
else:
annot_var = None
self.ann_hidden_warning.setVisible(
self.bar_size < 5 and annot_var is not None)
if self._silplot is not None:
if annot_var is not None:
column, _ = self.data.get_column_view(annot_var)
if self._mask is not None:
assert column.shape == self._mask.shape
# pylint: disable=invalid-unary-operand-type
column = column[~self._mask]
self._silplot.setRowNames(
[annot_var.str_val(value) for value in column])
else:
self._silplot.setRowNames(None)
def _update_scene_rect(self):
self.scene.setSceneRect(self._silplot.geometry())
def commit(self):
"""
Commit/send the current selection to the output.
"""
selected = indices = data = None
if self.data is not None:
selectedmask = np.full(len(self.data), False, dtype=bool)
if self._silplot is not None:
indices = self._silplot.selection()
assert (np.diff(indices) > 0).all(), "strictly increasing"
if self._mask is not None:
# pylint: disable=invalid-unary-operand-type
indices = np.flatnonzero(~self._mask)[indices]
selectedmask[indices] = True
if self._mask is not None:
scores = np.full(shape=selectedmask.shape,
fill_value=np.nan)
# pylint: disable=invalid-unary-operand-type
scores[~self._mask] = self._silhouette
else:
scores = self._silhouette
silhouette_var = None
if self.add_scores:
var = self.cluster_var_model[self.cluster_var_idx]
silhouette_var = Orange.data.ContinuousVariable(
"Silhouette ({})".format(escape(var.name)))
domain = Orange.data.Domain(
self.data.domain.attributes,
self.data.domain.class_vars,
self.data.domain.metas + (silhouette_var, ))
data = self.data.transform(domain)
else:
domain = self.data.domain
data = self.data
if np.count_nonzero(selectedmask):
selected = self.data.from_table(
domain, self.data, np.flatnonzero(selectedmask))
if self.add_scores:
if selected is not None:
selected[:, silhouette_var] = np.c_[scores[selectedmask]]
data[:, silhouette_var] = np.c_[scores]
self.Outputs.selected_data.send(selected)
self.Outputs.annotated_data.send(create_annotated_table(data, indices))
def send_report(self):
if not len(self.cluster_var_model):
return
self.report_plot()
caption = "Silhouette plot ({} distance), clustered by '{}'".format(
self.Distances[self.distance_idx][0],
self.cluster_var_model[self.cluster_var_idx])
if self.annotation_var_idx and self._silplot.rowNamesVisible():
caption += ", annotated with '{}'".format(
self.annotation_var_model[self.annotation_var_idx])
self.report_caption(caption)
def onDeleteWidget(self):
self.clear()
super().onDeleteWidget()
class OWBoxPlot(widget.OWWidget):
"""
Here's how the widget's functions call each other:
- `set_data` is a signal handler fills the list boxes and calls
`attr_changed`.
- `attr_changed` handles changes of attribute or grouping (callbacks for
list boxes). It recomputes box data by calling `compute_box_data`, shows
the appropriate display box (discrete/continuous) and then calls
`layout_changed`
- `layout_changed` constructs all the elements for the scene (as lists of
QGraphicsItemGroup) and calls `display_changed`. It is called when the
attribute or grouping is changed (by attr_changed) and on resize event.
- `display_changed` puts the elements corresponding to the current display
settings on the scene. It is called when the elements are reconstructed
(layout is changed due to selection of attributes or resize event), or
when the user changes display settings or colors.
For discrete attributes, the flow is a bit simpler: the elements are not
constructed in advance (by layout_changed). Instead, layout_changed and
display_changed call display_changed_disc that draws everything.
"""
name = "Box Plot"
description = "Visualize the distribution of feature values in a box plot."
icon = "icons/BoxPlot.svg"
priority = 100
inputs = [("Data", Orange.data.Table, "set_data")]
#: Comparison types for continuous variables
CompareNone, CompareMedians, CompareMeans = 0, 1, 2
settingsHandler = DomainContextHandler()
attribute = ContextSetting(None)
group_var = ContextSetting(None)
show_annotations = Setting(True)
compare = Setting(CompareMedians)
stattest = Setting(0)
sig_threshold = Setting(0.05)
stretched = Setting(True)
_sorting_criteria_attrs = {
CompareNone: "", CompareMedians: "median", CompareMeans: "mean"
}
_pen_axis_tick = QPen(Qt.white, 5)
_pen_axis = QPen(Qt.darkGray, 3)
_pen_median = QPen(QBrush(QColor(0xff, 0xff, 0x00)), 2)
_pen_paramet = QPen(QBrush(QColor(0x33, 0x00, 0xff)), 2)
_pen_dotted = QPen(QBrush(QColor(0x33, 0x00, 0xff)), 1)
_pen_dotted.setStyle(Qt.DotLine)
_post_line_pen = QPen(Qt.lightGray, 2)
_post_grp_pen = QPen(Qt.lightGray, 4)
for pen in (_pen_paramet, _pen_median, _pen_dotted,
_pen_axis, _pen_axis_tick, _post_line_pen, _post_grp_pen):
pen.setCosmetic(True)
pen.setCapStyle(Qt.RoundCap)
pen.setJoinStyle(Qt.RoundJoin)
_pen_axis_tick.setCapStyle(Qt.FlatCap)
_box_brush = QBrush(QColor(0x33, 0x88, 0xff, 0xc0))
_axis_font = QFont()
_axis_font.setPixelSize(12)
_label_font = QFont()
_label_font.setPixelSize(11)
_attr_brush = QBrush(QColor(0x33, 0x00, 0xff))
graph_name = "box_scene"
def __init__(self):
super().__init__()
self.stats = []
self.dataset = None
self.posthoc_lines = []
self.label_txts = self.mean_labels = self.boxes = self.labels = \
self.label_txts_all = self.attr_labels = self.order = []
self.p = -1.0
self.scale_x = self.scene_min_x = self.scene_width = 0
self.label_width = 0
common_options = dict(
callback=self.attr_changed, sizeHint=(200, 100))
self.attrs = VariableListModel()
gui.listView(
self.controlArea, self, "attribute", box="Variable",
model=self.attrs, **common_options)
self.group_vars = VariableListModel()
gui.listView(
self.controlArea, self, "group_var", box="Grouping",
model=self.group_vars, **common_options)
# TODO: move Compare median/mean to grouping box
self.display_box = gui.vBox(self.controlArea, "Display")
gui.checkBox(self.display_box, self, "show_annotations", "Annotate",
callback=self.display_changed)
self.compare_rb = gui.radioButtonsInBox(
self.display_box, self, 'compare',
btnLabels=["No comparison", "Compare medians", "Compare means"],
callback=self.display_changed)
self.stretching_box = gui.checkBox(
self.controlArea, self, 'stretched', "Stretch bars", box='Display',
callback=self.display_changed).box
gui.vBox(self.mainArea, addSpace=True)
self.box_scene = QGraphicsScene()
self.box_view = QGraphicsView(self.box_scene)
self.box_view.setRenderHints(QPainter.Antialiasing |
QPainter.TextAntialiasing |
QPainter.SmoothPixmapTransform)
self.box_view.viewport().installEventFilter(self)
self.mainArea.layout().addWidget(self.box_view)
e = gui.hBox(self.mainArea, addSpace=False)
self.infot1 = gui.widgetLabel(e, "<center>No test results.</center>")
self.mainArea.setMinimumWidth(650)
self.stats = self.dist = self.conts = []
self.is_continuous = False
self.update_display_box()
def eventFilter(self, obj, event):
if obj is self.box_view.viewport() and \
event.type() == QEvent.Resize:
self.layout_changed()
return super().eventFilter(obj, event)
# noinspection PyTypeChecker
def set_data(self, dataset):
if dataset is not None and (
not bool(dataset) or not len(dataset.domain)):
dataset = None
self.closeContext()
self.dataset = dataset
self.dist = self.stats = self.conts = []
self.group_var = None
self.attribute = None
if dataset:
domain = dataset.domain
self.group_vars[:] = \
[None] + \
[a for a in chain(domain.variables, domain.metas)
if a.is_discrete]
self.attrs[:] = chain(domain.variables,
(a for a in domain.metas if a.is_primitive()))
if self.attrs:
self.attribute = self.attrs[0]
if domain.class_var and domain.class_var.is_discrete:
self.group_var = domain.class_var
else:
self.group_var = None # Reset to trigger selection via callback
self.openContext(self.dataset)
self.attr_changed()
else:
self.reset_all_data()
def reset_all_data(self):
self.clear_scene()
self.infot1.setText("")
self.attrs[:] = []
self.group_vars[:] = []
self.is_continuous = False
self.update_display_box()
def attr_changed(self):
self.compute_box_data()
self.update_display_box()
self.layout_changed()
if self.is_continuous:
heights = 90 if self.show_annotations else 60
self.box_view.centerOn(self.scene_min_x + self.scene_width / 2,
-30 - len(self.stats) * heights / 2 + 45)
else:
self.box_view.centerOn(self.scene_width / 2,
-30 - len(self.boxes) * 40 / 2 + 45)
def compute_box_data(self):
attr = self.attribute
if not attr:
return
dataset = self.dataset
if dataset is None:
self.stats = self.dist = self.conts = []
return
self.is_continuous = attr.is_continuous
if self.group_var:
self.dist = []
self.conts = contingency.get_contingency(
dataset, attr, self.group_var)
if self.is_continuous:
self.stats = [BoxData(cont) for cont in self.conts]
self.label_txts_all = self.group_var.values
else:
self.dist = distribution.get_distribution(dataset, attr)
self.conts = []
if self.is_continuous:
self.stats = [BoxData(self.dist)]
self.label_txts_all = [""]
self.label_txts = [txts for stat, txts in zip(self.stats,
self.label_txts_all)
if stat.n > 0]
self.stats = [stat for stat in self.stats if stat.n > 0]
def update_display_box(self):
if self.is_continuous:
self.stretching_box.hide()
self.display_box.show()
self.compare_rb.setEnabled(self.group_var is not None)
else:
self.stretching_box.show()
self.display_box.hide()
def clear_scene(self):
self.box_scene.clear()
self.attr_labels = []
self.labels = []
self.boxes = []
self.mean_labels = []
self.posthoc_lines = []
def layout_changed(self):
attr = self.attribute
if not attr:
return
self.clear_scene()
if self.dataset is None or len(self.conts) == len(self.dist) == 0:
return
if not self.is_continuous:
return self.display_changed_disc()
self.mean_labels = [self.mean_label(stat, attr, lab)
for stat, lab in zip(self.stats, self.label_txts)]
self.draw_axis()
self.boxes = [self.box_group(stat) for stat in self.stats]
self.labels = [self.label_group(stat, attr, mean_lab)
for stat, mean_lab in zip(self.stats, self.mean_labels)]
self.attr_labels = [QGraphicsSimpleTextItem(lab)
for lab in self.label_txts]
for it in chain(self.labels, self.boxes, self.attr_labels):
self.box_scene.addItem(it)
self.display_changed()
def display_changed(self):
if self.dataset is None:
return
if not self.is_continuous:
return self.display_changed_disc()
self.order = list(range(len(self.stats)))
criterion = self._sorting_criteria_attrs[self.compare]
if criterion:
self.order = sorted(
self.order, key=lambda i: getattr(self.stats[i], criterion))
heights = 90 if self.show_annotations else 60
for row, box_index in enumerate(self.order):
y = (-len(self.stats) + row) * heights + 10
self.boxes[box_index].setY(y)
labels = self.labels[box_index]
if self.show_annotations:
labels.show()
labels.setY(y)
else:
labels.hide()
label = self.attr_labels[box_index]
label.setY(y - 15 - label.boundingRect().height())
if self.show_annotations:
label.hide()
else:
stat = self.stats[box_index]
if self.compare == OWBoxPlot.CompareMedians:
pos = stat.median + 5 / self.scale_x
elif self.compare == OWBoxPlot.CompareMeans:
pos = stat.mean + 5 / self.scale_x
else:
pos = stat.q25
label.setX(pos * self.scale_x)
label.show()
r = QRectF(self.scene_min_x, -30 - len(self.stats) * heights,
self.scene_width, len(self.stats) * heights + 90)
self.box_scene.setSceneRect(r)
self.compute_tests()
self.show_posthoc()
def display_changed_disc(self):
self.clear_scene()
self.attr_labels = [QGraphicsSimpleTextItem(lab)
for lab in self.label_txts_all]
if not self.stretched:
if self.group_var:
self.labels = [
QGraphicsTextItem("{}".format(int(sum(cont))))
for cont in self.conts]
else:
self.labels = [
QGraphicsTextItem(str(int(sum(self.dist))))]
self.draw_axis_disc()
if self.group_var:
self.boxes = [self.strudel(cont) for cont in self.conts]
else:
self.boxes = [self.strudel(self.dist)]
for row, box in enumerate(self.boxes):
y = (-len(self.boxes) + row) * 40 + 10
label = self.attr_labels[row]
b = label.boundingRect()
label.setPos(-b.width() - 10, y - b.height() / 2)
self.box_scene.addItem(label)
if not self.stretched:
label = self.labels[row]
b = label.boundingRect()
if self.group_var:
right = self.scale_x * sum(self.conts[row])
else:
right = self.scale_x * sum(self.dist)
label.setPos(right + 10, y - b.height() / 2)
self.box_scene.addItem(label)
if self.attribute is not self.group_var:
for text_item, bar_part in zip(box.childItems()[1::2],
box.childItems()[::2]):
label = QGraphicsSimpleTextItem(
text_item.toPlainText())
label.setPos(bar_part.boundingRect().x(),
y - label.boundingRect().height() - 8)
self.box_scene.addItem(label)
for text_item in box.childItems()[1::2]:
box.removeFromGroup(text_item)
self.box_scene.addItem(box)
box.setPos(0, y)
self.box_scene.setSceneRect(-self.label_width - 5,
-30 - len(self.boxes) * 40,
self.scene_width, len(self.boxes * 40) + 90)
self.infot1.setText("")
# noinspection PyPep8Naming
def compute_tests(self):
# The t-test and ANOVA are implemented here since they efficiently use
# the widget-specific data in self.stats.
# The non-parametric tests can't do this, so we use statistics.tests
def stat_ttest():
d1, d2 = self.stats
pooled_var = d1.var / d1.n + d2.var / d2.n
df = pooled_var ** 2 / \
((d1.var / d1.n) ** 2 / (d1.n - 1) +
(d2.var / d2.n) ** 2 / (d2.n - 1))
t = abs(d1.mean - d2.mean) / math.sqrt(pooled_var)
p = 2 * (1 - scipy.special.stdtr(df, t))
return t, p
# TODO: Check this function
# noinspection PyPep8Naming
def stat_ANOVA():
n = sum(stat.n for stat in self.stats)
grand_avg = sum(stat.n * stat.mean for stat in self.stats) / n
var_between = sum(stat.n * (stat.mean - grand_avg) ** 2
for stat in self.stats)
df_between = len(self.stats) - 1
var_within = sum(stat.n * stat.var for stat in self.stats)
df_within = n - len(self.stats)
F = (var_between / df_between) / (var_within / df_within)
p = 1 - scipy.special.fdtr(df_between, df_within, F)
return F, p
if self.compare == OWBoxPlot.CompareNone or len(self.stats) < 2:
t = ""
elif any(s.n <= 1 for s in self.stats):
t = "At least one group has just one instance, " \
"cannot compute significance"
elif len(self.stats) == 2:
if self.compare == OWBoxPlot.CompareMedians:
t = ""
# z, self.p = tests.wilcoxon_rank_sum(
# self.stats[0].dist, self.stats[1].dist)
# t = "Mann-Whitney's z: %.1f (p=%.3f)" % (z, self.p)
else:
t, self.p = stat_ttest()
t = "Student's t: %.3f (p=%.3f)" % (t, self.p)
else:
if self.compare == OWBoxPlot.CompareMedians:
t = ""
# U, self.p = -1, -1
# t = "Kruskal Wallis's U: %.1f (p=%.3f)" % (U, self.p)
else:
F, self.p = stat_ANOVA()
t = "ANOVA: %.3f (p=%.3f)" % (F, self.p)
self.infot1.setText("<center>%s</center>" % t)
def mean_label(self, stat, attr, val_name):
label = QGraphicsItemGroup()
t = QGraphicsSimpleTextItem(
"%.*f" % (attr.number_of_decimals + 1, stat.mean), label)
t.setFont(self._label_font)
bbox = t.boundingRect()
w2, h = bbox.width() / 2, bbox.height()
t.setPos(-w2, -h)
tpm = QGraphicsSimpleTextItem(
" \u00b1 " + "%.*f" % (attr.number_of_decimals + 1, stat.dev),
label)
tpm.setFont(self._label_font)
tpm.setPos(w2, -h)
if val_name:
vnm = QGraphicsSimpleTextItem(val_name + ": ", label)
vnm.setFont(self._label_font)
vnm.setBrush(self._attr_brush)
vb = vnm.boundingRect()
label.min_x = -w2 - vb.width()
vnm.setPos(label.min_x, -h)
else:
label.min_x = -w2
return label
def draw_axis(self):
"""Draw the horizontal axis and sets self.scale_x"""
misssing_stats = not self.stats
stats = self.stats or [BoxData(np.array([[0.], [1.]]))]
mean_labels = self.mean_labels or [self.mean_label(stats[0], self.attribute, "")]
bottom = min(stat.a_min for stat in stats)
top = max(stat.a_max for stat in stats)
first_val, step = compute_scale(bottom, top)
while bottom <= first_val:
first_val -= step
bottom = first_val
no_ticks = math.ceil((top - first_val) / step) + 1
top = max(top, first_val + no_ticks * step)
gbottom = min(bottom, min(stat.mean - stat.dev for stat in stats))
gtop = max(top, max(stat.mean + stat.dev for stat in stats))
bv = self.box_view
viewrect = bv.viewport().rect().adjusted(15, 15, -15, -30)
self.scale_x = scale_x = viewrect.width() / (gtop - gbottom)
# In principle we should repeat this until convergence since the new
# scaling is too conservative. (No chance am I doing this.)
mlb = min(stat.mean + mean_lab.min_x / scale_x
for stat, mean_lab in zip(stats, mean_labels))
if mlb < gbottom:
gbottom = mlb
self.scale_x = scale_x = viewrect.width() / (gtop - gbottom)
self.scene_min_x = gbottom * scale_x
self.scene_width = (gtop - gbottom) * scale_x
val = first_val
while True:
l = self.box_scene.addLine(val * scale_x, -1, val * scale_x, 1,
self._pen_axis_tick)
l.setZValue(100)
t = self.box_scene.addSimpleText(
self.attribute.repr_val(val) if not misssing_stats else "?",
self._axis_font)
t.setFlags(
t.flags() | QGraphicsItem.ItemIgnoresTransformations)
r = t.boundingRect()
t.setPos(val * scale_x - r.width() / 2, 8)
if val >= top:
break
val += step
self.box_scene.addLine(
bottom * scale_x - 4, 0, top * scale_x + 4, 0, self._pen_axis)
def draw_axis_disc(self):
"""
Draw the horizontal axis and sets self.scale_x for discrete attributes
"""
if self.stretched:
step = steps = 10
else:
if self.group_var:
max_box = max(float(np.sum(dist)) for dist in self.conts)
else:
max_box = float(np.sum(self.dist))
if max_box == 0:
self.scale_x = 1
return
_, step = compute_scale(0, max_box)
step = int(step) if step > 1 else 1
steps = int(math.ceil(max_box / step))
max_box = step * steps
bv = self.box_view
viewrect = bv.viewport().rect().adjusted(15, 15, -15, -30)
self.scene_width = viewrect.width()
lab_width = max(lab.boundingRect().width() for lab in self.attr_labels)
lab_width = max(lab_width, 40)
lab_width = min(lab_width, self.scene_width / 3)
self.label_width = lab_width
right_offset = 0 # offset for the right label
if not self.stretched and self.labels:
if self.group_var:
rows = list(zip(self.conts, self.labels))
else:
rows = [(self.dist, self.labels[0])]
# available space left of the 'group labels'
available = self.scene_width - lab_width - 10
scale_x = (available - right_offset) / max_box
max_right = max(sum(dist) * scale_x + 10 +
lbl.boundingRect().width()
for dist, lbl in rows)
right_offset = max(0, max_right - max_box * scale_x)
self.scale_x = scale_x = \
(self.scene_width - lab_width - 10 - right_offset) / max_box
self.box_scene.addLine(0, 0, max_box * scale_x, 0, self._pen_axis)
for val in range(0, step * steps + 1, step):
l = self.box_scene.addLine(val * scale_x, -1, val * scale_x, 1,
self._pen_axis_tick)
l.setZValue(100)
t = self.box_scene.addSimpleText(str(val), self._axis_font)
t.setPos(val * scale_x - t.boundingRect().width() / 2, 8)
if self.stretched:
self.scale_x *= 100
def label_group(self, stat, attr, mean_lab):
def centered_text(val, pos):
t = QGraphicsSimpleTextItem(
"%.*f" % (attr.number_of_decimals + 1, val), labels)
t.setFont(self._label_font)
bbox = t.boundingRect()
t.setPos(pos - bbox.width() / 2, 22)
return t
def line(x, down=1):
QGraphicsLineItem(x, 12 * down, x, 20 * down, labels)
def move_label(label, frm, to):
label.setX(to)
to += t_box.width() / 2
path = QPainterPath()
path.lineTo(0, 4)
path.lineTo(to - frm, 4)
path.lineTo(to - frm, 8)
p = QGraphicsPathItem(path)
p.setPos(frm, 12)
labels.addToGroup(p)
labels = QGraphicsItemGroup()
labels.addToGroup(mean_lab)
m = stat.mean * self.scale_x
mean_lab.setPos(m, -22)
line(m, -1)
msc = stat.median * self.scale_x
med_t = centered_text(stat.median, msc)
med_box_width2 = med_t.boundingRect().width()
line(msc)
x = stat.q25 * self.scale_x
t = centered_text(stat.q25, x)
t_box = t.boundingRect()
med_left = msc - med_box_width2
if x + t_box.width() / 2 >= med_left - 5:
move_label(t, x, med_left - t_box.width() - 5)
else:
line(x)
x = stat.q75 * self.scale_x
t = centered_text(stat.q75, x)
t_box = t.boundingRect()
med_right = msc + med_box_width2
if x - t_box.width() / 2 <= med_right + 5:
move_label(t, x, med_right + 5)
else:
line(x)
return labels
def box_group(self, stat, height=20):
def line(x0, y0, x1, y1, *args):
return QGraphicsLineItem(x0 * scale_x, y0, x1 * scale_x, y1, *args)
scale_x = self.scale_x
box = QGraphicsItemGroup()
whisker1 = line(stat.a_min, -1.5, stat.a_min, 1.5, box)
whisker2 = line(stat.a_max, -1.5, stat.a_max, 1.5, box)
vert_line = line(stat.a_min, 0, stat.a_max, 0, box)
mean_line = line(stat.mean, -height / 3, stat.mean, height / 3, box)
for it in (whisker1, whisker2, mean_line):
it.setPen(self._pen_paramet)
vert_line.setPen(self._pen_dotted)
var_line = line(stat.mean - stat.dev, 0, stat.mean + stat.dev, 0, box)
var_line.setPen(self._pen_paramet)
mbox = QGraphicsRectItem(stat.q25 * scale_x, -height / 2,
(stat.q75 - stat.q25) * scale_x, height,
box)
mbox.setBrush(self._box_brush)
mbox.setPen(QPen(Qt.NoPen))
mbox.setZValue(-200)
median_line = line(stat.median, -height / 2,
stat.median, height / 2, box)
median_line.setPen(self._pen_median)
median_line.setZValue(-150)
return box
def strudel(self, dist):
attr = self.attribute
ss = np.sum(dist)
box = QGraphicsItemGroup()
if ss < 1e-6:
QGraphicsRectItem(0, -10, 1, 10, box)
cum = 0
for i, v in enumerate(dist):
if v < 1e-6:
continue
if self.stretched:
v /= ss
v *= self.scale_x
rect = QGraphicsRectItem(cum + 1, -6, v - 2, 12, box)
rect.setBrush(QBrush(QColor(*attr.colors[i])))
rect.setPen(QPen(Qt.NoPen))
if self.stretched:
tooltip = "{}: {:.2f}%".format(attr.values[i],
100 * dist[i] / sum(dist))
else:
tooltip = "{}: {}".format(attr.values[i], int(dist[i]))
rect.setToolTip(tooltip)
text = QGraphicsTextItem(attr.values[i])
box.addToGroup(text)
cum += v
return box
def show_posthoc(self):
def line(y0, y1):
it = self.box_scene.addLine(x, y0, x, y1, self._post_line_pen)
it.setZValue(-100)
self.posthoc_lines.append(it)
while self.posthoc_lines:
self.box_scene.removeItem(self.posthoc_lines.pop())
if self.compare == OWBoxPlot.CompareNone or len(self.stats) < 2:
return
if self.compare == OWBoxPlot.CompareMedians:
crit_line = "median"
else:
crit_line = "mean"
xs = []
height = 90 if self.show_annotations else 60
y_up = -len(self.stats) * height + 10
for pos, box_index in enumerate(self.order):
stat = self.stats[box_index]
x = getattr(stat, crit_line) * self.scale_x
xs.append(x)
by = y_up + pos * height
line(by + 12, 3)
line(by - 12, by - 25)
used_to = []
last_to = to = 0
for frm, frm_x in enumerate(xs[:-1]):
for to in range(frm + 1, len(xs)):
if xs[to] - frm_x > 1.5:
to -= 1
break
if last_to == to or frm == to:
continue
for rowi, used in enumerate(used_to):
if used < frm:
used_to[rowi] = to
break
else:
rowi = len(used_to)
used_to.append(to)
y = - 6 - rowi * 6
it = self.box_scene.addLine(frm_x - 2, y, xs[to] + 2, y,
self._post_grp_pen)
self.posthoc_lines.append(it)
last_to = to
def get_widget_name_extension(self):
if self.attribute:
return self.attribute.name
def send_report(self):
self.report_plot()
text = ""
if self.attribute:
text += "Box plot for attribute '{}' ".format(self.attribute.name)
if self.group_var:
text += "grouped by '{}'".format(self.group_var.name)
if text:
self.report_caption(text)
class OWPythagoreanForest(OWWidget):
name = 'Pythagorean Forest'
description = 'Pythagorean forest for visualising random forests.'
icon = 'icons/PythagoreanForest.svg'
priority = 1001
inputs = [('Random forest', RandomForestModel, 'set_rf')]
outputs = [('Tree', TreeModel)]
# Enable the save as feature
graph_name = 'scene'
# Settings
depth_limit = settings.ContextSetting(10)
target_class_index = settings.ContextSetting(0)
size_calc_idx = settings.Setting(0)
zoom = settings.Setting(50)
selected_tree_index = settings.ContextSetting(-1)
def __init__(self):
super().__init__()
self.model = None
self.forest_adapter = None
self.instances = None
self.clf_dataset = None
# We need to store refernces to the trees and grid items
self.grid_items, self.ptrees = [], []
# In some rare cases, we need to prevent commiting, the only one
# that this currently helps is that when changing the size calculation
# the trees are all recomputed, but we don't want to output a new tree
# to keep things consistent with other ui controls.
self.__prevent_commit = False
self.color_palette = None
# Different methods to calculate the size of squares
self.SIZE_CALCULATION = [
('Normal', lambda x: x),
('Square root', lambda x: sqrt(x)),
('Logarithmic', lambda x: log(x + 1)),
]
# CONTROL AREA
# Tree info area
box_info = gui.widgetBox(self.controlArea, 'Forest')
self.ui_info = gui.widgetLabel(box_info)
# Display controls area
box_display = gui.widgetBox(self.controlArea, 'Display')
self.ui_depth_slider = gui.hSlider(
box_display, self, 'depth_limit', label='Depth', ticks=False,
callback=self.update_depth)
self.ui_target_class_combo = gui.comboBox(
box_display, self, 'target_class_index', label='Target class',
orientation=Qt.Horizontal, items=[], contentsLength=8,
callback=self.update_colors)
self.ui_size_calc_combo = gui.comboBox(
box_display, self, 'size_calc_idx', label='Size',
orientation=Qt.Horizontal,
items=list(zip(*self.SIZE_CALCULATION))[0], contentsLength=8,
callback=self.update_size_calc)
self.ui_zoom_slider = gui.hSlider(
box_display, self, 'zoom', label='Zoom', ticks=False, minValue=20,
maxValue=150, callback=self.zoom_changed, createLabel=False)
# Stretch to fit the rest of the unsused area
gui.rubber(self.controlArea)
self.controlArea.setSizePolicy(QSizePolicy.Preferred, QSizePolicy.Expanding)
# MAIN AREA
self.scene = QGraphicsScene(self)
self.scene.selectionChanged.connect(self.commit)
self.grid = OWGrid()
self.grid.geometryChanged.connect(self._update_scene_rect)
self.scene.addItem(self.grid)
self.view = QGraphicsView(self.scene)
self.view.setRenderHint(QPainter.Antialiasing, True)
self.view.setVerticalScrollBarPolicy(Qt.ScrollBarAlwaysOn)
self.mainArea.layout().addWidget(self.view)
self.resize(800, 500)
self.clear()
def set_rf(self, model=None):
"""When a different forest is given."""
self.clear()
self.model = model
if model is not None:
self.forest_adapter = self._get_forest_adapter(self.model)
self._draw_trees()
self.color_palette = self.forest_adapter.get_trees()[0]
self.instances = model.instances
# this bit is important for the regression classifier
if self.instances is not None and self.instances.domain != model.domain:
self.clf_dataset = self.instances.transform(self.model.domain)
else:
self.clf_dataset = self.instances
self._update_info_box()
self._update_target_class_combo()
self._update_depth_slider()
self.selected_tree_index = -1
def clear(self):
"""Clear all relevant data from the widget."""
self.model = None
self.forest_adapter = None
self.ptrees = []
self.grid_items = []
self.grid.clear()
self._clear_info_box()
self._clear_target_class_combo()
self._clear_depth_slider()
def update_depth(self):
"""When the max depth slider is changed."""
for tree in self.ptrees:
tree.set_depth_limit(self.depth_limit)
def update_colors(self):
"""When the target class or coloring method is changed."""
for tree in self.ptrees:
tree.target_class_changed(self.target_class_index)
def update_size_calc(self):
"""When the size calculation of the trees is changed."""
if self.model is not None:
with self._prevent_commit():
self.grid.clear()
self._draw_trees()
# Keep the selected item
if self.selected_tree_index != -1:
self.grid_items[self.selected_tree_index].setSelected(True)
self.update_depth()
def zoom_changed(self):
"""When we update the "Zoom" slider."""
for item in self.grid_items:
item.set_max_size(self._calculate_zoom(self.zoom))
width = (self.view.width() - self.view.verticalScrollBar().width())
self.grid.reflow(width)
self.grid.setPreferredWidth(width)
@contextmanager
def _prevent_commit(self):
try:
self.__prevent_commit = True
yield
finally:
self.__prevent_commit = False
def _update_info_box(self):
self.ui_info.setText('Trees: {}'.format(len(self.forest_adapter.get_trees())))
def _update_depth_slider(self):
self.depth_limit = self._get_max_depth()
self.ui_depth_slider.parent().setEnabled(True)
self.ui_depth_slider.setMaximum(self.depth_limit)
self.ui_depth_slider.setValue(self.depth_limit)
def _clear_info_box(self):
self.ui_info.setText('No forest on input.')
def _clear_target_class_combo(self):
self.ui_target_class_combo.clear()
self.target_class_index = 0
self.ui_target_class_combo.setCurrentIndex(self.target_class_index)
def _clear_depth_slider(self):
self.ui_depth_slider.parent().setEnabled(False)
self.ui_depth_slider.setMaximum(0)
def _get_max_depth(self):
return max(tree.tree_adapter.max_depth for tree in self.ptrees)
def _get_forest_adapter(self, model):
return SklRandomForestAdapter(model)
@contextmanager
def disable_ui(self):
"""Temporarly disable the UI while trees may be redrawn."""
try:
self.ui_size_calc_combo.setEnabled(False)
self.ui_depth_slider.setEnabled(False)
self.ui_target_class_combo.setEnabled(False)
self.ui_zoom_slider.setEnabled(False)
yield
finally:
self.ui_size_calc_combo.setEnabled(True)
self.ui_depth_slider.setEnabled(True)
self.ui_target_class_combo.setEnabled(True)
self.ui_zoom_slider.setEnabled(True)
def _draw_trees(self):
self.grid_items, self.ptrees = [], []
num_trees = len(self.forest_adapter.get_trees())
with self.progressBar(num_trees) as prg, self.disable_ui():
for tree in self.forest_adapter.get_trees():
ptree = PythagorasTreeViewer(
None, tree, interactive=False, padding=100,
target_class_index=self.target_class_index,
weight_adjustment=self.SIZE_CALCULATION[self.size_calc_idx][1]
)
grid_item = GridItem(
ptree, self.grid, max_size=self._calculate_zoom(self.zoom)
)
# We don't want to show flickering while the trees are being
grid_item.setVisible(False)
self.grid_items.append(grid_item)
self.ptrees.append(ptree)
prg.advance()
self.grid.set_items(self.grid_items)
# This is necessary when adding items for the first time
if self.grid:
width = (self.view.width() - self.view.verticalScrollBar().width())
self.grid.reflow(width)
self.grid.setPreferredWidth(width)
# After drawing is complete, we show the trees
for grid_item in self.grid_items:
grid_item.setVisible(True)
@staticmethod
def _calculate_zoom(zoom_level):
"""Calculate the max size for grid items from zoom level setting."""
return zoom_level * 5
def onDeleteWidget(self):
"""When deleting the widget."""
super().onDeleteWidget()
self.clear()
def commit(self):
"""Commit the selected tree to output."""
if self.__prevent_commit:
return
if not self.scene.selectedItems():
self.send('Tree', None)
# The selected tree index should only reset when model changes
if self.model is None:
self.selected_tree_index = -1
return
selected_item = self.scene.selectedItems()[0]
self.selected_tree_index = self.grid_items.index(selected_item)
tree = self.model.trees[self.selected_tree_index]
tree.instances = self.instances
tree.meta_target_class_index = self.target_class_index
tree.meta_size_calc_idx = self.size_calc_idx
tree.meta_depth_limit = self.depth_limit
self.send('Tree', tree)
def send_report(self):
"""Send report."""
self.report_plot()
def _update_scene_rect(self):
self.scene.setSceneRect(self.scene.itemsBoundingRect())
def _update_target_class_combo(self):
self._clear_target_class_combo()
label = [x for x in self.ui_target_class_combo.parent().children()
if isinstance(x, QLabel)][0]
if self.instances.domain.has_discrete_class:
label_text = 'Target class'
values = [c.title() for c in self.instances.domain.class_vars[0].values]
values.insert(0, 'None')
else:
label_text = 'Node color'
values = list(ContinuousTreeNode.COLOR_METHODS.keys())
label.setText(label_text)
self.ui_target_class_combo.addItems(values)
self.ui_target_class_combo.setCurrentIndex(self.target_class_index)
def resizeEvent(self, ev):
width = (self.view.width() - self.view.verticalScrollBar().width())
self.grid.reflow(width)
self.grid.setPreferredWidth(width)
super().resizeEvent(ev)
class OWMosaicDisplay(OWWidget):
name = "Mosaic Display"
description = "Display data in a mosaic plot."
icon = "icons/MosaicDisplay.svg"
priority = 220
keywords = []
class Inputs:
data = Input("Data", Table, default=True)
data_subset = Input("Data Subset", Table)
class Outputs:
selected_data = Output("Selected Data", Table, default=True)
annotated_data = Output(ANNOTATED_DATA_SIGNAL_NAME, Table)
settingsHandler = DomainContextHandler()
vizrank = SettingProvider(MosaicVizRank)
settings_version = 2
use_boxes = Setting(True)
variable1 = ContextSetting(None)
variable2 = ContextSetting(None)
variable3 = ContextSetting(None)
variable4 = ContextSetting(None)
variable_color = ContextSetting(None)
selection = ContextSetting(set())
BAR_WIDTH = 5
SPACING = 4
ATTR_NAME_OFFSET = 20
ATTR_VAL_OFFSET = 3
BLUE_COLORS = [QColor(255, 255, 255), QColor(210, 210, 255),
QColor(110, 110, 255), QColor(0, 0, 255)]
RED_COLORS = [QColor(255, 255, 255), QColor(255, 200, 200),
QColor(255, 100, 100), QColor(255, 0, 0)]
graph_name = "canvas"
attrs_changed_manually = Signal(list)
class Warning(OWWidget.Warning):
incompatible_subset = Msg("Data subset is incompatible with Data")
no_valid_data = Msg("No valid data")
no_cont_selection_sql = \
Msg("Selection of numeric features on SQL is not supported")
def __init__(self):
super().__init__()
self.data = None
self.discrete_data = None
self.subset_data = None
self.subset_indices = None
self.color_data = None
self.areas = []
self.canvas = QGraphicsScene()
self.canvas_view = ViewWithPress(
self.canvas, handler=self.clear_selection)
self.mainArea.layout().addWidget(self.canvas_view)
self.canvas_view.setVerticalScrollBarPolicy(Qt.ScrollBarAlwaysOff)
self.canvas_view.setHorizontalScrollBarPolicy(Qt.ScrollBarAlwaysOff)
self.canvas_view.setRenderHint(QPainter.Antialiasing)
box = gui.vBox(self.controlArea, box=True)
self.model_1 = DomainModel(
order=DomainModel.MIXED, valid_types=DomainModel.PRIMITIVE)
self.model_234 = DomainModel(
order=DomainModel.MIXED, valid_types=DomainModel.PRIMITIVE,
placeholder="(None)")
self.attr_combos = [
gui.comboBox(
box, self, value="variable{}".format(i),
orientation=Qt.Horizontal, contentsLength=12,
callback=self.attr_changed,
model=self.model_1 if i == 1 else self.model_234)
for i in range(1, 5)]
self.vizrank, self.vizrank_button = MosaicVizRank.add_vizrank(
box, self, "Find Informative Mosaics", self.set_attr)
box2 = gui.vBox(self.controlArea, box="Interior Coloring")
self.color_model = DomainModel(
order=DomainModel.MIXED, valid_types=DomainModel.PRIMITIVE,
placeholder="(Pearson residuals)")
self.cb_attr_color = gui.comboBox(
box2, self, value="variable_color",
orientation=Qt.Horizontal, contentsLength=12, labelWidth=50,
callback=self.set_color_data, model=self.color_model)
self.bar_button = gui.checkBox(
box2, self, 'use_boxes', label='Compare with total',
callback=self.update_graph)
gui.rubber(self.controlArea)
def sizeHint(self):
return QSize(720, 530)
def _get_discrete_data(self, data):
"""
Discretize continuous attributes.
Return None when there is no data, no rows, or no primitive attributes.
"""
if (data is None or
not len(data) or
not any(attr.is_discrete or attr.is_continuous
for attr in chain(data.domain.variables,
data.domain.metas))):
return None
elif any(attr.is_continuous for attr in data.domain.variables):
return Discretize(
method=EqualFreq(n=4), remove_const=False, discretize_classes=True,
discretize_metas=True)(data)
else:
return data
def init_combos(self, data):
def set_combos(value):
self.model_1.set_domain(value)
self.model_234.set_domain(value)
self.color_model.set_domain(value)
if data is None:
set_combos(None)
self.variable1 = self.variable2 = self.variable3 \
= self.variable4 = self.variable_color = None
return
set_combos(self.data.domain)
if len(self.model_1) > 0:
self.variable1 = self.model_1[0]
self.variable2 = self.model_1[min(1, len(self.model_1) - 1)]
self.variable3 = self.variable4 = None
self.variable_color = self.data.domain.class_var # None is OK, too
def get_disc_attr_list(self):
return [self.discrete_data.domain[var.name]
for var in (self.variable1, self.variable2,
self.variable3, self.variable4)
if var]
def set_attr(self, *attrs):
self.variable1, self.variable2, self.variable3, self.variable4 = [
attr and self.data.domain[attr.name] for attr in attrs]
self.reset_graph()
def attr_changed(self):
self.attrs_changed_manually.emit(self.get_disc_attr_list())
self.reset_graph()
def resizeEvent(self, e):
OWWidget.resizeEvent(self, e)
self.update_graph()
def showEvent(self, ev):
OWWidget.showEvent(self, ev)
self.update_graph()
@Inputs.data
def set_data(self, data):
if isinstance(data, SqlTable) and data.approx_len() > LARGE_TABLE:
data = data.sample_time(DEFAULT_SAMPLE_TIME)
self.closeContext()
self.data = data
self.vizrank.stop_and_reset()
self.vizrank_button.setEnabled(
self.data is not None and len(self.data) > 1
and len(self.data.domain.attributes) >= 1)
if self.data is None:
self.discrete_data = None
self.init_combos(None)
return
self.init_combos(self.data)
self.openContext(self.data)
@Inputs.data_subset
def set_subset_data(self, data):
self.subset_data = data
# this is called by widget after setData and setSubsetData are called.
# this way the graph is updated only once
def handleNewSignals(self):
self.Warning.incompatible_subset.clear()
self.subset_indices = None
if self.data is not None and self.subset_data:
transformed = self.subset_data.transform(self.data.domain)
if np.all(np.isnan(transformed.X)) \
and np.all(np.isnan(transformed.Y)):
self.Warning.incompatible_subset()
else:
indices = {e.id for e in transformed}
self.subset_indices = [ex.id in indices for ex in self.data]
self.set_color_data()
self.reset_graph()
def clear_selection(self):
self.selection = set()
self.update_selection_rects()
self.send_selection()
def coloring_changed(self):
self.vizrank.coloring_changed()
self.update_graph()
def reset_graph(self):
self.clear_selection()
self.update_graph()
def set_color_data(self):
if self.data is None:
return
self.bar_button.setEnabled(self.variable_color is not None)
attrs = [v for v in self.model_1 if v and v is not self.variable_color]
domain = Domain(attrs, self.variable_color, None)
self.color_data = self.data.from_table(domain, self.data)
self.discrete_data = self._get_discrete_data(self.color_data)
self.vizrank.stop_and_reset()
self.vizrank_button.setEnabled(True)
self.coloring_changed()
def update_selection_rects(self):
pens = (QPen(), QPen(Qt.black, 3, Qt.DotLine))
for i, (_, _, area) in enumerate(self.areas):
area.setPen(pens[i in self.selection])
def select_area(self, index, ev):
if ev.button() != Qt.LeftButton:
return
if ev.modifiers() & Qt.ControlModifier:
self.selection ^= {index}
else:
self.selection = {index}
self.update_selection_rects()
self.send_selection()
def send_selection(self):
if not self.selection or self.data is None:
self.Outputs.selected_data.send(None)
self.Outputs.annotated_data.send(
create_annotated_table(self.data, []))
return
filters = []
self.Warning.no_cont_selection_sql.clear()
if self.discrete_data is not self.data:
if isinstance(self.data, SqlTable):
self.Warning.no_cont_selection_sql()
for i in self.selection:
cols, vals, _ = self.areas[i]
filters.append(
filter.Values(
filter.FilterDiscrete(col, [val])
for col, val in zip(cols, vals)))
if len(filters) > 1:
filters = filter.Values(filters, conjunction=False)
else:
filters = filters[0]
selection = filters(self.discrete_data)
idset = set(selection.ids)
sel_idx = [i for i, id in enumerate(self.data.ids) if id in idset]
if self.discrete_data is not self.data:
selection = self.data[sel_idx]
self.Outputs.selected_data.send(selection)
self.Outputs.annotated_data.send(
create_annotated_table(self.data, sel_idx))
def send_report(self):
self.report_plot(self.canvas)
def update_graph(self):
spacing = self.SPACING
bar_width = self.BAR_WIDTH
def get_counts(attr_vals, values):
"""Calculate rectangles' widths; if all are 0, they are set to 1."""
if not attr_vals:
counts = [conditionaldict[val] for val in values]
else:
counts = [conditionaldict[attr_vals + "-" + val]
for val in values]
total = sum(counts)
if total == 0:
counts = [1] * len(values)
total = sum(counts)
return total, counts
def draw_data(attr_list, x0_x1, y0_y1, side, condition,
total_attrs, used_attrs, used_vals, attr_vals=""):
x0, x1 = x0_x1
y0, y1 = y0_y1
if conditionaldict[attr_vals] == 0:
add_rect(x0, x1, y0, y1, "",
used_attrs, used_vals, attr_vals=attr_vals)
# store coordinates for later drawing of labels
draw_text(side, attr_list[0], (x0, x1), (y0, y1), total_attrs,
used_attrs, used_vals, attr_vals)
return
attr = attr_list[0]
# how much smaller rectangles do we draw
edge = len(attr_list) * spacing
values = get_variable_values_sorted(attr)
if side % 2:
values = values[::-1] # reverse names if necessary
if side % 2 == 0: # we are drawing on the x axis
# remove the space needed for separating different attr. values
whole = max(0, (x1 - x0) - edge * (len(values) - 1))
if whole == 0:
edge = (x1 - x0) / float(len(values) - 1)
else: # we are drawing on the y axis
whole = max(0, (y1 - y0) - edge * (len(values) - 1))
if whole == 0:
edge = (y1 - y0) / float(len(values) - 1)
total, counts = get_counts(attr_vals, values)
# when visualizing the third attribute and the first attribute has
# the last value, reverse the order in which the boxes are drawn;
# otherwise, if the last cell, nearest to the labels of the fourth
# attribute, is empty, we wouldn't be able to position the labels
valrange = list(range(len(values)))
if len(attr_list + used_attrs) == 4 and len(used_attrs) == 2:
attr1values = get_variable_values_sorted(used_attrs[0])
if used_vals[0] == attr1values[-1]:
valrange = valrange[::-1]
for i in valrange:
start = i * edge + whole * float(sum(counts[:i]) / total)
end = i * edge + whole * float(sum(counts[:i + 1]) / total)
val = values[i]
htmlval = to_html(val)
newattrvals = attr_vals + "-" + val if attr_vals else val
tooltip = "{} {}: <b>{}</b><br/>".format(
condition, attr.name, htmlval)
attrs = used_attrs + [attr]
vals = used_vals + [val]
args = attrs, vals, newattrvals
if side % 2 == 0: # if we are moving horizontally
if len(attr_list) == 1:
add_rect(x0 + start, x0 + end, y0, y1, tooltip, *args)
else:
draw_data(
attr_list[1:], (x0 + start, x0 + end), (y0, y1),
side + 1, tooltip, total_attrs, *args)
else:
if len(attr_list) == 1:
add_rect(x0, x1, y0 + start, y0 + end, tooltip, *args)
else:
draw_data(
attr_list[1:], (x0, x1), (y0 + start, y0 + end),
side + 1, tooltip, total_attrs, *args)
draw_text(side, attr_list[0], (x0, x1), (y0, y1),
total_attrs, used_attrs, used_vals, attr_vals)
def draw_text(side, attr, x0_x1, y0_y1,
total_attrs, used_attrs, used_vals, attr_vals):
x0, x1 = x0_x1
y0, y1 = y0_y1
if side in drawn_sides:
return
# the text on the right will be drawn when we are processing
# visualization of the last value of the first attribute
if side == 3:
attr1values = get_variable_values_sorted(used_attrs[0])
if used_vals[0] != attr1values[-1]:
return
if not conditionaldict[attr_vals]:
if side not in draw_positions:
draw_positions[side] = (x0, x1, y0, y1)
return
else:
if side in draw_positions:
# restore the positions of attribute values and name
(x0, x1, y0, y1) = draw_positions[side]
drawn_sides.add(side)
values = get_variable_values_sorted(attr)
if side % 2:
values = values[::-1]
spaces = spacing * (total_attrs - side) * (len(values) - 1)
width = x1 - x0 - spaces * (side % 2 == 0)
height = y1 - y0 - spaces * (side % 2 == 1)
# calculate position of first attribute
currpos = 0
total, counts = get_counts(attr_vals, values)
aligns = [Qt.AlignTop | Qt.AlignHCenter,
Qt.AlignRight | Qt.AlignVCenter,
Qt.AlignBottom | Qt.AlignHCenter,
Qt.AlignLeft | Qt.AlignVCenter]
align = aligns[side]
for i, val in enumerate(values):
if distributiondict[val] != 0:
perc = counts[i] / float(total)
xs = [x0 + currpos + width * 0.5 * perc,
x0 - self.ATTR_VAL_OFFSET,
x0 + currpos + width * perc * 0.5,
x1 + self.ATTR_VAL_OFFSET]
ys = [y1 + self.ATTR_VAL_OFFSET,
y0 + currpos + height * 0.5 * perc,
y0 - self.ATTR_VAL_OFFSET,
y0 + currpos + height * 0.5 * perc]
CanvasText(self.canvas, val, xs[side], ys[side], align)
space = height if side % 2 else width
currpos += perc * space + spacing * (total_attrs - side)
xs = [x0 + (x1 - x0) / 2,
x0 - max_ylabel_w1 - self.ATTR_VAL_OFFSET,
x0 + (x1 - x0) / 2,
x1 + max_ylabel_w2 + self.ATTR_VAL_OFFSET]
ys = [y1 + self.ATTR_VAL_OFFSET + self.ATTR_NAME_OFFSET,
y0 + (y1 - y0) / 2,
y0 - self.ATTR_VAL_OFFSET - self.ATTR_NAME_OFFSET,
y0 + (y1 - y0) / 2]
CanvasText(
self.canvas, attr.name, xs[side], ys[side], align, bold=True,
vertical=side % 2)
def add_rect(x0, x1, y0, y1, condition,
used_attrs, used_vals, attr_vals=""):
area_index = len(self.areas)
x1 += (x0 == x1)
y1 += (y0 == y1)
# rectangles of width and height 1 are not shown - increase
y1 += (x1 - x0 + y1 - y0 == 2)
colors = class_var and [QColor(*col) for col in class_var.colors]
def select_area(_, ev):
self.select_area(area_index, ev)
def rect(x, y, w, h, z, pen_color=None, brush_color=None, **args):
if pen_color is None:
return CanvasRectangle(
self.canvas, x, y, w, h, z=z, onclick=select_area,
**args)
if brush_color is None:
brush_color = pen_color
return CanvasRectangle(
self.canvas, x, y, w, h, pen_color, brush_color, z=z,
onclick=select_area, **args)
def line(x1, y1, x2, y2):
r = QGraphicsLineItem(x1, y1, x2, y2, None)
self.canvas.addItem(r)
r.setPen(QPen(Qt.white, 2))
r.setZValue(30)
outer_rect = rect(x0, y0, x1 - x0, y1 - y0, 30)
self.areas.append((used_attrs, used_vals, outer_rect))
if not conditionaldict[attr_vals]:
return
if self.variable_color is None:
s = sum(apriori_dists[0])
expected = s * reduce(
mul,
(apriori_dists[i][used_vals[i]] / float(s)
for i in range(len(used_vals))))
actual = conditionaldict[attr_vals]
pearson = float((actual - expected) / sqrt(expected))
if pearson == 0:
ind = 0
else:
ind = max(0, min(int(log(abs(pearson), 2)), 3))
color = [self.RED_COLORS, self.BLUE_COLORS][pearson > 0][ind]
rect(x0, y0, x1 - x0, y1 - y0, -20, color)
outer_rect.setToolTip(
condition + "<hr/>" +
"Expected instances: %.1f<br>"
"Actual instances: %d<br>"
"Standardized (Pearson) residual: %.1f" %
(expected, conditionaldict[attr_vals], pearson))
else:
cls_values = get_variable_values_sorted(class_var)
prior = get_distribution(data, class_var.name)
total = 0
for i, value in enumerate(cls_values):
val = conditionaldict[attr_vals + "-" + value]
if val == 0:
continue
if i == len(cls_values) - 1:
v = y1 - y0 - total
else:
v = ((y1 - y0) * val) / conditionaldict[attr_vals]
rect(x0, y0 + total, x1 - x0, v, -20, colors[i])
total += v
if self.use_boxes and \
abs(x1 - x0) > bar_width and abs(y1 - y0) > bar_width:
total = 0
line(x0 + bar_width, y0, x0 + bar_width, y1)
n = sum(prior)
for i, (val, color) in enumerate(zip(prior, colors)):
if i == len(prior) - 1:
h = y1 - y0 - total
else:
h = (y1 - y0) * val / n
rect(x0, y0 + total, bar_width, h, 20, color)
total += h
if conditionalsubsetdict:
if conditionalsubsetdict[attr_vals]:
if self.subset_indices is not None:
line(x1 - bar_width, y0, x1 - bar_width, y1)
total = 0
n = conditionalsubsetdict[attr_vals]
if n:
for i, (cls, color) in \
enumerate(zip(cls_values, colors)):
val = conditionalsubsetdict[
attr_vals + "-" + cls]
if val == 0:
continue
if i == len(prior) - 1:
v = y1 - y0 - total
else:
v = ((y1 - y0) * val) / n
rect(x1 - bar_width, y0 + total,
bar_width, v, 15, color)
total += v
actual = [conditionaldict[attr_vals + "-" + cls_values[i]]
for i in range(len(prior))]
n_actual = sum(actual)
if n_actual > 0:
apriori = [prior[key] for key in cls_values]
n_apriori = sum(apriori)
text = "<br/>".join(
"<b>%s</b>: %d / %.1f%% (Expected %.1f / %.1f%%)" %
(cls, act, 100.0 * act / n_actual,
apr / n_apriori * n_actual, 100.0 * apr / n_apriori)
for cls, act, apr in zip(cls_values, actual, apriori))
else:
text = ""
outer_rect.setToolTip(
"{}<hr>Instances: {}<br><br>{}".format(
condition, n_actual, text[:-4]))
def draw_legend(x0_x1, y0_y1):
x0, x1 = x0_x1
_, y1 = y0_y1
if self.variable_color is None:
names = ["<-8", "-8:-4", "-4:-2", "-2:2", "2:4", "4:8", ">8",
"Residuals:"]
colors = self.RED_COLORS[::-1] + self.BLUE_COLORS[1:]
else:
names = get_variable_values_sorted(class_var) + \
[class_var.name + ":"]
colors = [QColor(*col) for col in class_var.colors]
names = [CanvasText(self.canvas, name, alignment=Qt.AlignVCenter)
for name in names]
totalwidth = sum(text.boundingRect().width() for text in names)
# compute the x position of the center of the legend
y = y1 + self.ATTR_NAME_OFFSET + self.ATTR_VAL_OFFSET + 35
distance = 30
startx = (x0 + x1) / 2 - (totalwidth + (len(names)) * distance) / 2
names[-1].setPos(startx + 15, y)
names[-1].show()
xoffset = names[-1].boundingRect().width() + distance
size = 8
for i in range(len(names) - 1):
if self.variable_color is None:
edgecolor = Qt.black
else:
edgecolor = colors[i]
CanvasRectangle(self.canvas, startx + xoffset, y - size / 2,
size, size, edgecolor, colors[i])
names[i].setPos(startx + xoffset + 10, y)
xoffset += distance + names[i].boundingRect().width()
self.canvas.clear()
self.areas = []
data = self.discrete_data
if data is None:
return
attr_list = self.get_disc_attr_list()
class_var = data.domain.class_var
if class_var:
sql = isinstance(data, SqlTable)
name = not sql and data.name
# save class_var because it is removed in the next line
data = data[:, attr_list + [class_var]]
data.domain.class_var = class_var
if not sql:
data.name = name
else:
data = data[:, attr_list]
# TODO: check this
# data = Preprocessor_dropMissing(data)
if len(data) == 0:
self.Warning.no_valid_data()
return
else:
self.Warning.no_valid_data.clear()
attrs = [attr for attr in attr_list if not attr.values]
if attrs:
CanvasText(self.canvas,
"Feature {} has no values".format(attrs[0]),
(self.canvas_view.width() - 120) / 2,
self.canvas_view.height() / 2)
return
if self.variable_color is None:
apriori_dists = [get_distribution(data, attr) for attr in attr_list]
else:
apriori_dists = []
def get_max_label_width(attr):
values = get_variable_values_sorted(attr)
maxw = 0
for val in values:
t = CanvasText(self.canvas, val, 0, 0, bold=0, show=False)
maxw = max(int(t.boundingRect().width()), maxw)
return maxw
# get the maximum width of rectangle
xoff = 20
width = 20
max_ylabel_w1 = max_ylabel_w2 = 0
if len(attr_list) > 1:
text = CanvasText(self.canvas, attr_list[1].name, bold=1, show=0)
max_ylabel_w1 = min(get_max_label_width(attr_list[1]), 150)
width = 5 + text.boundingRect().height() + \
self.ATTR_VAL_OFFSET + max_ylabel_w1
xoff = width
if len(attr_list) == 4:
text = CanvasText(self.canvas, attr_list[3].name, bold=1, show=0)
max_ylabel_w2 = min(get_max_label_width(attr_list[3]), 150)
width += text.boundingRect().height() + \
self.ATTR_VAL_OFFSET + max_ylabel_w2 - 10
# get the maximum height of rectangle
height = 100
yoff = 45
square_size = min(self.canvas_view.width() - width - 20,
self.canvas_view.height() - height - 20)
if square_size < 0:
return # canvas is too small to draw rectangles
self.canvas_view.setSceneRect(
0, 0, self.canvas_view.width(), self.canvas_view.height())
drawn_sides = set()
draw_positions = {}
conditionaldict, distributiondict = \
get_conditional_distribution(data, attr_list)
conditionalsubsetdict = None
if self.subset_indices:
conditionalsubsetdict, _ = get_conditional_distribution(
self.discrete_data[self.subset_indices], attr_list)
# draw rectangles
draw_data(
attr_list, (xoff, xoff + square_size), (yoff, yoff + square_size),
0, "", len(attr_list), [], [])
draw_legend((xoff, xoff + square_size), (yoff, yoff + square_size))
self.update_selection_rects()
@classmethod
def migrate_context(cls, context, version):
if version < 2:
settings.migrate_str_to_variable(context, none_placeholder="(None)")
class OWBoxPlot(widget.OWWidget):
"""
Here's how the widget's functions call each other:
- `set_data` is a signal handler fills the list boxes and calls
`grouping_changed`.
- `grouping_changed` handles changes of grouping attribute: it enables or
disables the box for ordering, orders attributes and calls `attr_changed`.
- `attr_changed` handles changes of attribute. It recomputes box data by
calling `compute_box_data`, shows the appropriate display box
(discrete/continuous) and then calls`layout_changed`
- `layout_changed` constructs all the elements for the scene (as lists of
QGraphicsItemGroup) and calls `display_changed`. It is called when the
attribute or grouping is changed (by attr_changed) and on resize event.
- `display_changed` puts the elements corresponding to the current display
settings on the scene. It is called when the elements are reconstructed
(layout is changed due to selection of attributes or resize event), or
when the user changes display settings or colors.
For discrete attributes, the flow is a bit simpler: the elements are not
constructed in advance (by layout_changed). Instead, layout_changed and
display_changed call display_changed_disc that draws everything.
"""
name = "Box Plot"
description = "Visualize the distribution of feature values in a box plot."
icon = "icons/BoxPlot.svg"
priority = 100
keywords = ["whisker"]
class Inputs:
data = Input("Data", Orange.data.Table)
class Outputs:
selected_data = Output("Selected Data", Orange.data.Table, default=True)
annotated_data = Output(ANNOTATED_DATA_SIGNAL_NAME, Orange.data.Table)
#: Comparison types for continuous variables
CompareNone, CompareMedians, CompareMeans = 0, 1, 2
settingsHandler = DomainContextHandler()
conditions = ContextSetting([])
attribute = ContextSetting(None)
order_by_importance = Setting(False)
group_var = ContextSetting(None)
show_annotations = Setting(True)
compare = Setting(CompareMeans)
stattest = Setting(0)
sig_threshold = Setting(0.05)
stretched = Setting(True)
show_labels = Setting(True)
sort_freqs = Setting(False)
auto_commit = Setting(True)
_sorting_criteria_attrs = {
CompareNone: "", CompareMedians: "median", CompareMeans: "mean"
}
_pen_axis_tick = QPen(Qt.white, 5)
_pen_axis = QPen(Qt.darkGray, 3)
_pen_median = QPen(QBrush(QColor(0xff, 0xff, 0x00)), 2)
_pen_paramet = QPen(QBrush(QColor(0x33, 0x00, 0xff)), 2)
_pen_dotted = QPen(QBrush(QColor(0x33, 0x00, 0xff)), 1)
_pen_dotted.setStyle(Qt.DotLine)
_post_line_pen = QPen(Qt.lightGray, 2)
_post_grp_pen = QPen(Qt.lightGray, 4)
for pen in (_pen_paramet, _pen_median, _pen_dotted,
_pen_axis, _pen_axis_tick, _post_line_pen, _post_grp_pen):
pen.setCosmetic(True)
pen.setCapStyle(Qt.RoundCap)
pen.setJoinStyle(Qt.RoundJoin)
_pen_axis_tick.setCapStyle(Qt.FlatCap)
_box_brush = QBrush(QColor(0x33, 0x88, 0xff, 0xc0))
_axis_font = QFont()
_axis_font.setPixelSize(12)
_label_font = QFont()
_label_font.setPixelSize(11)
_attr_brush = QBrush(QColor(0x33, 0x00, 0xff))
graph_name = "box_scene"
def __init__(self):
super().__init__()
self.stats = []
self.dataset = None
self.posthoc_lines = []
self.label_txts = self.mean_labels = self.boxes = self.labels = \
self.label_txts_all = self.attr_labels = self.order = []
self.scale_x = self.scene_min_x = self.scene_width = 0
self.label_width = 0
self.attrs = VariableListModel()
view = gui.listView(
self.controlArea, self, "attribute", box="Variable",
model=self.attrs, callback=self.attr_changed)
view.setMinimumSize(QSize(30, 30))
# Any other policy than Ignored will let the QListBox's scrollbar
# set the minimal height (see the penultimate paragraph of
# http://doc.qt.io/qt-4.8/qabstractscrollarea.html#addScrollBarWidget)
view.setSizePolicy(QSizePolicy.Expanding, QSizePolicy.Ignored)
gui.separator(view.box, 6, 6)
self.cb_order = gui.checkBox(
view.box, self, "order_by_importance",
"Order by relevance",
tooltip="Order by 𝜒² or ANOVA over the subgroups",
callback=self.apply_sorting)
self.group_vars = DomainModel(
placeholder="None", separators=False,
valid_types=Orange.data.DiscreteVariable)
self.group_view = view = gui.listView(
self.controlArea, self, "group_var", box="Subgroups",
model=self.group_vars, callback=self.grouping_changed)
view.setEnabled(False)
view.setMinimumSize(QSize(30, 30))
# See the comment above
view.setSizePolicy(QSizePolicy.Expanding, QSizePolicy.Ignored)
# TODO: move Compare median/mean to grouping box
# The vertical size policy is needed to let only the list views expand
self.display_box = gui.vBox(
self.controlArea, "Display",
sizePolicy=(QSizePolicy.Minimum, QSizePolicy.Maximum),
addSpace=False)
gui.checkBox(self.display_box, self, "show_annotations", "Annotate",
callback=self.display_changed)
self.compare_rb = gui.radioButtonsInBox(
self.display_box, self, 'compare',
btnLabels=["No comparison", "Compare medians", "Compare means"],
callback=self.layout_changed)
# The vertical size policy is needed to let only the list views expand
self.stretching_box = box = gui.vBox(
self.controlArea, box="Display",
sizePolicy=(QSizePolicy.Minimum, QSizePolicy.Fixed))
self.stretching_box.sizeHint = self.display_box.sizeHint
gui.checkBox(
box, self, 'stretched', "Stretch bars",
callback=self.display_changed)
gui.checkBox(
box, self, 'show_labels', "Show box labels",
callback=self.display_changed)
self.sort_cb = gui.checkBox(
box, self, 'sort_freqs', "Sort by subgroup frequencies",
callback=self.display_changed)
gui.rubber(box)
gui.auto_commit(self.controlArea, self, "auto_commit",
"Send Selection", "Send Automatically")
gui.vBox(self.mainArea, addSpace=True)
self.box_scene = QGraphicsScene()
self.box_scene.selectionChanged.connect(self.commit)
self.box_view = QGraphicsView(self.box_scene)
self.box_view.setRenderHints(QPainter.Antialiasing |
QPainter.TextAntialiasing |
QPainter.SmoothPixmapTransform)
self.box_view.viewport().installEventFilter(self)
self.mainArea.layout().addWidget(self.box_view)
e = gui.hBox(self.mainArea, addSpace=False)
self.infot1 = gui.widgetLabel(e, "<center>No test results.</center>")
self.mainArea.setMinimumWidth(300)
self.stats = self.dist = self.conts = []
self.is_continuous = False
self.update_display_box()
def sizeHint(self):
return QSize(900, 500)
def eventFilter(self, obj, event):
if obj is self.box_view.viewport() and \
event.type() == QEvent.Resize:
self.layout_changed()
return super().eventFilter(obj, event)
def reset_attrs(self, domain):
self.attrs[:] = [
var for var in chain(
domain.class_vars, domain.metas, domain.attributes)
if var.is_primitive()]
# noinspection PyTypeChecker
@Inputs.data
def set_data(self, dataset):
if dataset is not None and (
not bool(dataset) or not len(dataset.domain) and not
any(var.is_primitive() for var in dataset.domain.metas)):
dataset = None
self.closeContext()
self.dataset = dataset
self.dist = self.stats = self.conts = []
self.group_var = None
self.attribute = None
if dataset:
domain = dataset.domain
self.group_vars.set_domain(domain)
self.group_view.setEnabled(len(self.group_vars) > 1)
self.reset_attrs(domain)
self.select_default_variables(domain)
self.openContext(self.dataset)
self.grouping_changed()
else:
self.reset_all_data()
self.commit()
def select_default_variables(self, domain):
# visualize first non-class variable, group by class (if present)
if len(self.attrs) > len(domain.class_vars):
self.attribute = self.attrs[len(domain.class_vars)]
elif self.attrs:
self.attribute = self.attrs[0]
if domain.class_var and domain.class_var.is_discrete:
self.group_var = domain.class_var
else:
self.group_var = None # Reset to trigger selection via callback
def apply_sorting(self):
def compute_score(attr):
if attr is group_var:
return 3
if attr.is_continuous:
# One-way ANOVA
col = data.get_column_view(attr)[0].astype(float)
groups = (col[group_col == i] for i in range(n_groups))
groups = (col[~np.isnan(col)] for col in groups)
groups = [group for group in groups if len(group)]
p = f_oneway(*groups)[1] if len(groups) > 1 else 2
else:
# Chi-square with the given distribution into groups
# (see degrees of freedom in computation of the p-value)
if not attr.values or not group_var.values:
return 2
observed = np.array(
contingency.get_contingency(data, group_var, attr))
observed = observed[observed.sum(axis=1) != 0, :]
observed = observed[:, observed.sum(axis=0) != 0]
if min(observed.shape) < 2:
return 2
expected = \
np.outer(observed.sum(axis=1), observed.sum(axis=0)) / \
np.sum(observed)
p = chisquare(observed.ravel(), f_exp=expected.ravel(),
ddof=n_groups - 1)[1]
if math.isnan(p):
return 2
return p
data = self.dataset
if data is None:
return
domain = data.domain
attribute = self.attribute
group_var = self.group_var
if self.order_by_importance and group_var is not None:
n_groups = len(group_var.values)
group_col = data.get_column_view(group_var)[0] if \
domain.has_continuous_attributes(
include_class=True, include_metas=True) else None
self.attrs.sort(key=compute_score)
else:
self.reset_attrs(domain)
self.attribute = attribute
def reset_all_data(self):
self.clear_scene()
self.infot1.setText("")
self.attrs.clear()
self.group_vars.set_domain(None)
self.group_view.setEnabled(False)
self.is_continuous = False
self.update_display_box()
def grouping_changed(self):
self.cb_order.setEnabled(self.group_var is not None)
self.apply_sorting()
self.attr_changed()
def select_box_items(self):
temp_cond = self.conditions.copy()
for box in self.box_scene.items():
if isinstance(box, FilterGraphicsRectItem):
box.setSelected(box.filter.conditions in
[c.conditions for c in temp_cond])
def attr_changed(self):
self.compute_box_data()
self.update_display_box()
self.layout_changed()
if self.is_continuous:
heights = 90 if self.show_annotations else 60
self.box_view.centerOn(self.scene_min_x + self.scene_width / 2,
-30 - len(self.stats) * heights / 2 + 45)
else:
self.box_view.centerOn(self.scene_width / 2,
-30 - len(self.boxes) * 40 / 2 + 45)
def compute_box_data(self):
attr = self.attribute
if not attr:
return
dataset = self.dataset
self.is_continuous = attr.is_continuous
if dataset is None or not self.is_continuous and not attr.values or \
self.group_var and not self.group_var.values:
self.stats = self.dist = self.conts = []
return
if self.group_var:
self.dist = []
self.conts = contingency.get_contingency(
dataset, attr, self.group_var)
if self.is_continuous:
stats, label_texts = [], []
for i, cont in enumerate(self.conts):
if np.sum(cont[1]):
stats.append(BoxData(cont, attr, i, self.group_var))
label_texts.append(self.group_var.values[i])
self.stats = stats
self.label_txts_all = label_texts
else:
self.label_txts_all = \
[v for v, c in zip(self.group_var.values, self.conts)
if np.sum(c) > 0]
else:
self.dist = distribution.get_distribution(dataset, attr)
self.conts = []
if self.is_continuous:
self.stats = [BoxData(self.dist, attr, None)]
self.label_txts_all = [""]
self.label_txts = [txts for stat, txts in zip(self.stats,
self.label_txts_all)
if stat.n > 0]
self.stats = [stat for stat in self.stats if stat.n > 0]
def update_display_box(self):
if self.is_continuous:
self.stretching_box.hide()
self.display_box.show()
self.compare_rb.setEnabled(self.group_var is not None)
else:
self.stretching_box.show()
self.display_box.hide()
self.sort_cb.setEnabled(self.group_var is not None)
def clear_scene(self):
self.closeContext()
self.box_scene.clearSelection()
self.box_scene.clear()
self.box_view.viewport().update()
self.attr_labels = []
self.labels = []
self.boxes = []
self.mean_labels = []
self.posthoc_lines = []
self.openContext(self.dataset)
def layout_changed(self):
attr = self.attribute
if not attr:
return
self.clear_scene()
if self.dataset is None or len(self.conts) == len(self.dist) == 0:
return
if not self.is_continuous:
self.display_changed_disc()
return
self.mean_labels = [self.mean_label(stat, attr, lab)
for stat, lab in zip(self.stats, self.label_txts)]
self.draw_axis()
self.boxes = [self.box_group(stat) for stat in self.stats]
self.labels = [self.label_group(stat, attr, mean_lab)
for stat, mean_lab in zip(self.stats, self.mean_labels)]
self.attr_labels = [QGraphicsSimpleTextItem(lab)
for lab in self.label_txts]
for it in chain(self.labels, self.attr_labels):
self.box_scene.addItem(it)
self.display_changed()
def display_changed(self):
if self.dataset is None:
return
if not self.is_continuous:
self.display_changed_disc()
return
self.order = list(range(len(self.stats)))
criterion = self._sorting_criteria_attrs[self.compare]
if criterion:
vals = [getattr(stat, criterion) for stat in self.stats]
overmax = max((val for val in vals if val is not None), default=0) \
+ 1
vals = [val if val is not None else overmax for val in vals]
self.order = sorted(self.order, key=vals.__getitem__)
heights = 90 if self.show_annotations else 60
for row, box_index in enumerate(self.order):
y = (-len(self.stats) + row) * heights + 10
for item in self.boxes[box_index]:
self.box_scene.addItem(item)
item.setY(y)
labels = self.labels[box_index]
if self.show_annotations:
labels.show()
labels.setY(y)
else:
labels.hide()
label = self.attr_labels[box_index]
label.setY(y - 15 - label.boundingRect().height())
if self.show_annotations:
label.hide()
else:
stat = self.stats[box_index]
if self.compare == OWBoxPlot.CompareMedians and \
stat.median is not None:
pos = stat.median + 5 / self.scale_x
elif self.compare == OWBoxPlot.CompareMeans or stat.q25 is None:
pos = stat.mean + 5 / self.scale_x
else:
pos = stat.q25
label.setX(pos * self.scale_x)
label.show()
r = QRectF(self.scene_min_x, -30 - len(self.stats) * heights,
self.scene_width, len(self.stats) * heights + 90)
self.box_scene.setSceneRect(r)
self.compute_tests()
self.show_posthoc()
self.select_box_items()
def display_changed_disc(self):
assert not self.is_continuous
self.clear_scene()
self.attr_labels = [QGraphicsSimpleTextItem(lab)
for lab in self.label_txts_all]
if not self.stretched:
if self.group_var:
self.labels = [
QGraphicsTextItem("{}".format(int(sum(cont))))
for cont in self.conts if np.sum(cont) > 0]
else:
self.labels = [
QGraphicsTextItem(str(int(sum(self.dist))))]
self.order = list(range(len(self.attr_labels)))
self.draw_axis_disc()
if self.group_var:
self.boxes = \
[self.strudel(cont, i) for i, cont in enumerate(self.conts)
if np.sum(cont) > 0]
self.conts = self.conts[np.sum(np.array(self.conts), axis=1) > 0]
if self.sort_freqs:
# pylint: disable=invalid-unary-operand-type
self.order = sorted(self.order, key=(-np.sum(self.conts, axis=1)).__getitem__)
else:
self.boxes = [self.strudel(self.dist)]
for row, box_index in enumerate(self.order):
y = (-len(self.boxes) + row) * 40 + 10
box = self.boxes[box_index]
bars, labels = box[::2], box[1::2]
self.__draw_group_labels(y, box_index)
if not self.stretched:
self.__draw_row_counts(y, box_index)
if self.show_labels and self.attribute is not self.group_var:
self.__draw_bar_labels(y, bars, labels)
self.__draw_bars(y, bars)
self.box_scene.setSceneRect(-self.label_width - 5,
-30 - len(self.boxes) * 40,
self.scene_width, len(self.boxes * 40) + 90)
self.infot1.setText("")
self.select_box_items()
def __draw_group_labels(self, y, row):
"""Draw group labels
Parameters
----------
y: int
vertical offset of bars
row: int
row index
"""
label = self.attr_labels[row]
b = label.boundingRect()
label.setPos(-b.width() - 10, y - b.height() / 2)
self.box_scene.addItem(label)
def __draw_row_counts(self, y, row):
"""Draw row counts
Parameters
----------
y: int
vertical offset of bars
row: int
row index
"""
assert not self.is_continuous
label = self.labels[row]
b = label.boundingRect()
if self.group_var:
right = self.scale_x * sum(self.conts[row])
else:
right = self.scale_x * sum(self.dist)
label.setPos(right + 10, y - b.height() / 2)
self.box_scene.addItem(label)
def __draw_bar_labels(self, y, bars, labels):
"""Draw bar labels
Parameters
----------
y: int
vertical offset of bars
bars: List[FilterGraphicsRectItem]
list of bars being drawn
labels: List[QGraphicsTextItem]
list of labels for corresponding bars
"""
label = bar_part = None
for text_item, bar_part in zip(labels, bars):
label = self.Label(
text_item.toPlainText())
label.setPos(bar_part.boundingRect().x(),
y - label.boundingRect().height() - 8)
label.setMaxWidth(bar_part.boundingRect().width())
self.box_scene.addItem(label)
def __draw_bars(self, y, bars):
"""Draw bars
Parameters
----------
y: int
vertical offset of bars
bars: List[FilterGraphicsRectItem]
list of bars to draw
"""
for item in bars:
item.setPos(0, y)
self.box_scene.addItem(item)
# noinspection PyPep8Naming
def compute_tests(self):
# The t-test and ANOVA are implemented here since they efficiently use
# the widget-specific data in self.stats.
# The non-parametric tests can't do this, so we use statistics.tests
# pylint: disable=comparison-with-itself
def stat_ttest():
d1, d2 = self.stats
if d1.n < 2 or d2.n < 2:
return np.nan, np.nan
pooled_var = d1.var / d1.n + d2.var / d2.n
# pylint: disable=comparison-with-itself
if pooled_var == 0 or np.isnan(pooled_var):
return np.nan, np.nan
df = pooled_var ** 2 / \
((d1.var / d1.n) ** 2 / (d1.n - 1) +
(d2.var / d2.n) ** 2 / (d2.n - 1))
t = abs(d1.mean - d2.mean) / math.sqrt(pooled_var)
p = 2 * (1 - scipy.special.stdtr(df, t))
return t, p
# TODO: Check this function
# noinspection PyPep8Naming
def stat_ANOVA():
if any(stat.n == 0 for stat in self.stats):
return np.nan, np.nan
n = sum(stat.n for stat in self.stats)
grand_avg = sum(stat.n * stat.mean for stat in self.stats) / n
var_between = sum(stat.n * (stat.mean - grand_avg) ** 2
for stat in self.stats)
df_between = len(self.stats) - 1
var_within = sum(stat.n * stat.var for stat in self.stats)
df_within = n - len(self.stats)
if var_within == 0 or df_within == 0 or df_between == 0:
return np.nan, np.nan
F = (var_between / df_between) / (var_within / df_within)
p = 1 - scipy.special.fdtr(df_between, df_within, F)
return F, p
if self.compare == OWBoxPlot.CompareNone or len(self.stats) < 2:
t = ""
elif any(s.n <= 1 for s in self.stats):
t = "At least one group has just one instance, " \
"cannot compute significance"
elif len(self.stats) == 2:
if self.compare == OWBoxPlot.CompareMedians:
t = ""
# z, p = tests.wilcoxon_rank_sum(
# self.stats[0].dist, self.stats[1].dist)
# t = "Mann-Whitney's z: %.1f (p=%.3f)" % (z, p)
else:
t, p = stat_ttest()
t = "" if np.isnan(t) else f"Student's t: {t:.3f} (p={p:.3f})"
else:
if self.compare == OWBoxPlot.CompareMedians:
t = ""
# U, p = -1, -1
# t = "Kruskal Wallis's U: %.1f (p=%.3f)" % (U, p)
else:
F, p = stat_ANOVA()
t = "" if np.isnan(F) else f"ANOVA: {F:.3f} (p={p:.3f})"
self.infot1.setText("<center>%s</center>" % t)
def mean_label(self, stat, attr, val_name):
label = QGraphicsItemGroup()
t = QGraphicsSimpleTextItem(
"%.*f" % (attr.number_of_decimals + 1, stat.mean), label)
t.setFont(self._label_font)
bbox = t.boundingRect()
w2, h = bbox.width() / 2, bbox.height()
t.setPos(-w2, -h)
tpm = QGraphicsSimpleTextItem(
" \u00b1 " + "%.*f" % (attr.number_of_decimals + 1, stat.dev),
label)
tpm.setFont(self._label_font)
tpm.setPos(w2, -h)
if val_name:
vnm = QGraphicsSimpleTextItem(val_name + ": ", label)
vnm.setFont(self._label_font)
vnm.setBrush(self._attr_brush)
vb = vnm.boundingRect()
label.min_x = -w2 - vb.width()
vnm.setPos(label.min_x, -h)
else:
label.min_x = -w2
return label
def draw_axis(self):
"""Draw the horizontal axis and sets self.scale_x"""
misssing_stats = not self.stats
stats = self.stats or [BoxData(np.array([[0.], [1.]]), self.attribute)]
mean_labels = self.mean_labels or [self.mean_label(stats[0], self.attribute, "")]
bottom = min(stat.a_min for stat in stats)
top = max(stat.a_max for stat in stats)
first_val, step = compute_scale(bottom, top)
while bottom <= first_val:
first_val -= step
bottom = first_val
no_ticks = math.ceil((top - first_val) / step) + 1
top = max(top, first_val + no_ticks * step)
gbottom = min(bottom, min(stat.mean - stat.dev for stat in stats))
gtop = max(top, max(stat.mean + stat.dev for stat in stats))
bv = self.box_view
viewrect = bv.viewport().rect().adjusted(15, 15, -15, -30)
self.scale_x = scale_x = viewrect.width() / (gtop - gbottom)
# In principle we should repeat this until convergence since the new
# scaling is too conservative. (No chance am I doing this.)
mlb = min(stat.mean + mean_lab.min_x / scale_x
for stat, mean_lab in zip(stats, mean_labels))
if mlb < gbottom:
gbottom = mlb
self.scale_x = scale_x = viewrect.width() / (gtop - gbottom)
self.scene_min_x = gbottom * scale_x
self.scene_width = (gtop - gbottom) * scale_x
val = first_val
decimals = max(3, 4 - int(math.log10(step)))
while True:
l = self.box_scene.addLine(val * scale_x, -1, val * scale_x, 1,
self._pen_axis_tick)
l.setZValue(100)
t = self.box_scene.addSimpleText(
repr(round(val, decimals)) if not misssing_stats else "?",
self._axis_font)
t.setFlags(
t.flags() | QGraphicsItem.ItemIgnoresTransformations)
r = t.boundingRect()
t.setPos(val * scale_x - r.width() / 2, 8)
if val >= top:
break
val += step
self.box_scene.addLine(
bottom * scale_x - 4, 0, top * scale_x + 4, 0, self._pen_axis)
def draw_axis_disc(self):
"""
Draw the horizontal axis and sets self.scale_x for discrete attributes
"""
assert not self.is_continuous
if self.stretched:
if not self.attr_labels:
return
step = steps = 10
else:
if self.group_var:
max_box = max(float(np.sum(dist)) for dist in self.conts)
else:
max_box = float(np.sum(self.dist))
if max_box == 0:
self.scale_x = 1
return
_, step = compute_scale(0, max_box)
step = int(step) if step > 1 else 1
steps = int(math.ceil(max_box / step))
max_box = step * steps
bv = self.box_view
viewrect = bv.viewport().rect().adjusted(15, 15, -15, -30)
self.scene_width = viewrect.width()
lab_width = max(lab.boundingRect().width() for lab in self.attr_labels)
lab_width = max(lab_width, 40)
lab_width = min(lab_width, self.scene_width / 3)
self.label_width = lab_width
right_offset = 0 # offset for the right label
if not self.stretched and self.labels:
if self.group_var:
rows = list(zip(self.conts, self.labels))
else:
rows = [(self.dist, self.labels[0])]
# available space left of the 'group labels'
available = self.scene_width - lab_width - 10
scale_x = (available - right_offset) / max_box
max_right = max(sum(dist) * scale_x + 10 +
lbl.boundingRect().width()
for dist, lbl in rows)
right_offset = max(0, max_right - max_box * scale_x)
self.scale_x = scale_x = \
(self.scene_width - lab_width - 10 - right_offset) / max_box
self.box_scene.addLine(0, 0, max_box * scale_x, 0, self._pen_axis)
for val in range(0, step * steps + 1, step):
l = self.box_scene.addLine(val * scale_x, -1, val * scale_x, 1,
self._pen_axis_tick)
l.setZValue(100)
t = self.box_scene.addSimpleText(str(val), self._axis_font)
t.setPos(val * scale_x - t.boundingRect().width() / 2, 8)
if self.stretched:
self.scale_x *= 100
def label_group(self, stat, attr, mean_lab):
def centered_text(val, pos):
t = QGraphicsSimpleTextItem(
"%.*f" % (attr.number_of_decimals + 1, val), labels)
t.setFont(self._label_font)
bbox = t.boundingRect()
t.setPos(pos - bbox.width() / 2, 22)
return t
def line(x, down=1):
QGraphicsLineItem(x, 12 * down, x, 20 * down, labels)
def move_label(label, frm, to):
label.setX(to)
to += t_box.width() / 2
path = QPainterPath()
path.lineTo(0, 4)
path.lineTo(to - frm, 4)
path.lineTo(to - frm, 8)
p = QGraphicsPathItem(path)
p.setPos(frm, 12)
labels.addToGroup(p)
labels = QGraphicsItemGroup()
labels.addToGroup(mean_lab)
m = stat.mean * self.scale_x
mean_lab.setPos(m, -22)
line(m, -1)
if stat.median is not None:
msc = stat.median * self.scale_x
med_t = centered_text(stat.median, msc)
med_box_width2 = med_t.boundingRect().width() / 2
line(msc)
if stat.q25 is not None:
x = stat.q25 * self.scale_x
t = centered_text(stat.q25, x)
t_box = t.boundingRect()
med_left = msc - med_box_width2
if x + t_box.width() / 2 >= med_left - 5:
move_label(t, x, med_left - t_box.width() - 5)
else:
line(x)
if stat.q75 is not None:
x = stat.q75 * self.scale_x
t = centered_text(stat.q75, x)
t_box = t.boundingRect()
med_right = msc + med_box_width2
if x - t_box.width() / 2 <= med_right + 5:
move_label(t, x, med_right + 5)
else:
line(x)
return labels
def box_group(self, stat, height=20):
def line(x0, y0, x1, y1, *args):
return QGraphicsLineItem(x0 * scale_x, y0, x1 * scale_x, y1, *args)
scale_x = self.scale_x
box = []
whisker1 = line(stat.a_min, -1.5, stat.a_min, 1.5)
whisker2 = line(stat.a_max, -1.5, stat.a_max, 1.5)
vert_line = line(stat.a_min, 0, stat.a_max, 0)
mean_line = line(stat.mean, -height / 3, stat.mean, height / 3)
for it in (whisker1, whisker2, mean_line):
it.setPen(self._pen_paramet)
vert_line.setPen(self._pen_dotted)
var_line = line(stat.mean - stat.dev, 0, stat.mean + stat.dev, 0)
var_line.setPen(self._pen_paramet)
box.extend([whisker1, whisker2, vert_line, mean_line, var_line])
if stat.q25 is not None and stat.q75 is not None:
mbox = FilterGraphicsRectItem(
stat.conditions, stat.q25 * scale_x, -height / 2,
(stat.q75 - stat.q25) * scale_x, height)
mbox.setBrush(self._box_brush)
mbox.setPen(QPen(Qt.NoPen))
mbox.setZValue(-200)
box.append(mbox)
if stat.median is not None:
median_line = line(stat.median, -height / 2,
stat.median, height / 2)
median_line.setPen(self._pen_median)
median_line.setZValue(-150)
box.append(median_line)
return box
def strudel(self, dist, group_val_index=None):
attr = self.attribute
ss = np.sum(dist)
box = []
if ss < 1e-6:
cond = [FilterDiscrete(attr, None)]
if group_val_index is not None:
cond.append(FilterDiscrete(self.group_var, [group_val_index]))
box.append(FilterGraphicsRectItem(cond, 0, -10, 1, 10))
cum = 0
for i, v in enumerate(dist):
if v < 1e-6:
continue
if self.stretched:
v /= ss
v *= self.scale_x
cond = [FilterDiscrete(attr, [i])]
if group_val_index is not None:
cond.append(FilterDiscrete(self.group_var, [group_val_index]))
rect = FilterGraphicsRectItem(cond, cum + 1, -6, v - 2, 12)
rect.setBrush(QBrush(QColor(*attr.colors[i])))
rect.setPen(QPen(Qt.NoPen))
if self.stretched:
tooltip = "{}: {:.2f}%".format(attr.values[i],
100 * dist[i] / sum(dist))
else:
tooltip = "{}: {}".format(attr.values[i], int(dist[i]))
rect.setToolTip(tooltip)
text = QGraphicsTextItem(attr.values[i])
box.append(rect)
box.append(text)
cum += v
return box
def commit(self):
self.conditions = [item.filter for item in
self.box_scene.selectedItems() if item.filter]
selected, selection = None, []
if self.conditions:
selected = Values(self.conditions, conjunction=False)(self.dataset)
selection = np.in1d(
self.dataset.ids, selected.ids, assume_unique=True).nonzero()[0]
self.Outputs.selected_data.send(selected)
self.Outputs.annotated_data.send(
create_annotated_table(self.dataset, selection))
def show_posthoc(self):
def line(y0, y1):
it = self.box_scene.addLine(x, y0, x, y1, self._post_line_pen)
it.setZValue(-100)
self.posthoc_lines.append(it)
while self.posthoc_lines:
self.box_scene.removeItem(self.posthoc_lines.pop())
if self.compare == OWBoxPlot.CompareNone or len(self.stats) < 2:
return
if self.compare == OWBoxPlot.CompareMedians:
crit_line = "median"
else:
crit_line = "mean"
xs = []
height = 90 if self.show_annotations else 60
y_up = -len(self.stats) * height + 10
for pos, box_index in enumerate(self.order):
stat = self.stats[box_index]
x = getattr(stat, crit_line)
if x is None:
continue
x *= self.scale_x
xs.append(x * self.scale_x)
by = y_up + pos * height
line(by + 12, 3)
line(by - 12, by - 25)
used_to = []
last_to = to = 0
for frm, frm_x in enumerate(xs[:-1]):
for to in range(frm + 1, len(xs)):
if xs[to] - frm_x > 1.5:
to -= 1
break
if to in (last_to, frm):
continue
for rowi, used in enumerate(used_to):
if used < frm:
used_to[rowi] = to
break
else:
rowi = len(used_to)
used_to.append(to)
y = - 6 - rowi * 6
it = self.box_scene.addLine(frm_x - 2, y, xs[to] + 2, y,
self._post_grp_pen)
self.posthoc_lines.append(it)
last_to = to
def get_widget_name_extension(self):
return self.attribute.name if self.attribute else None
def send_report(self):
self.report_plot()
text = ""
if self.attribute:
text += "Box plot for attribute '{}' ".format(self.attribute.name)
if self.group_var:
text += "grouped by '{}'".format(self.group_var.name)
if text:
self.report_caption(text)
class Label(QGraphicsSimpleTextItem):
"""Boxplot Label with settable maxWidth"""
# Minimum width to display label text
MIN_LABEL_WIDTH = 25
# padding bellow the text
PADDING = 3
__max_width = None
def maxWidth(self):
return self.__max_width
def setMaxWidth(self, max_width):
self.__max_width = max_width
def paint(self, painter, option, widget):
"""Overrides QGraphicsSimpleTextItem.paint
If label text is too long, it is elided
to fit into the allowed region
"""
if self.__max_width is None:
width = option.rect.width()
else:
width = self.__max_width
if width < self.MIN_LABEL_WIDTH:
# if space is too narrow, no label
return
fm = painter.fontMetrics()
text = fm.elidedText(self.text(), Qt.ElideRight, width)
painter.drawText(
option.rect.x(),
option.rect.y() + self.boundingRect().height() - self.PADDING,
text)
class OWSilhouettePlot(widget.OWWidget):
name = "Silhouette Plot"
description = "Visually assess cluster quality and " \
"the degree of cluster membership."
icon = "icons/SilhouettePlot.svg"
priority = 300
inputs = [("Data", Orange.data.Table, "set_data")]
outputs = [("Selected Data", Orange.data.Table, widget.Default),
(ANNOTATED_DATA_SIGNAL_NAME, Orange.data.Table)]
replaces = [
"orangecontrib.prototypes.widgets.owsilhouetteplot.OWSilhouettePlot",
"Orange.widgets.unsupervised.owsilhouetteplot.OWSilhouettePlot"
]
settingsHandler = settings.PerfectDomainContextHandler()
#: Distance metric index
distance_idx = settings.Setting(0)
#: Group/cluster variable index
cluster_var_idx = settings.ContextSetting(0)
#: Annotation variable index
annotation_var_idx = settings.ContextSetting(0)
#: Group the silhouettes by cluster
group_by_cluster = settings.Setting(True)
#: A fixed size for an instance bar
bar_size = settings.Setting(3)
#: Add silhouette scores to output data
add_scores = settings.Setting(False)
auto_commit = settings.Setting(False)
Distances = [("Euclidean", Orange.distance.Euclidean),
("Manhattan", Orange.distance.Manhattan)]
graph_name = "scene"
buttons_area_orientation = Qt.Vertical
class Error(widget.OWWidget.Error):
need_two_clusters = Msg("Need at least two non-empty clusters")
def __init__(self):
super().__init__()
self.data = None
self._effective_data = None
self._matrix = None
self._silhouette = None
self._labels = None
self._silplot = None
gui.comboBox(
self.controlArea, self, "distance_idx", box="Distance",
items=[name for name, _ in OWSilhouettePlot.Distances],
orientation=Qt.Horizontal, callback=self._invalidate_distances)
box = gui.vBox(self.controlArea, "Cluster Label")
self.cluster_var_cb = gui.comboBox(
box, self, "cluster_var_idx", addSpace=4,
callback=self._invalidate_scores)
gui.checkBox(
box, self, "group_by_cluster", "Group by cluster",
callback=self._replot)
self.cluster_var_model = itemmodels.VariableListModel(parent=self)
self.cluster_var_cb.setModel(self.cluster_var_model)
box = gui.vBox(self.controlArea, "Bars")
gui.widgetLabel(box, "Bar width:")
gui.hSlider(
box, self, "bar_size", minValue=1, maxValue=10, step=1,
callback=self._update_bar_size, addSpace=6)
gui.widgetLabel(box, "Annotations:")
self.annotation_cb = gui.comboBox(
box, self, "annotation_var_idx", callback=self._update_annotations)
self.annotation_var_model = itemmodels.VariableListModel(parent=self)
self.annotation_var_model[:] = ["None"]
self.annotation_cb.setModel(self.annotation_var_model)
ibox = gui.indentedBox(box, 5)
self.ann_hidden_warning = warning = gui.widgetLabel(
ibox, "(increase the width to show)")
ibox.setFixedWidth(ibox.sizeHint().width())
warning.setVisible(False)
gui.rubber(self.controlArea)
gui.separator(self.buttonsArea)
box = gui.vBox(self.buttonsArea, "Output")
# Thunk the call to commit to call conditional commit
gui.checkBox(box, self, "add_scores", "Add silhouette scores",
callback=lambda: self.commit())
gui.auto_commit(
box, self, "auto_commit", "Commit",
auto_label="Auto commit", box=False)
# Ensure that the controlArea is not narrower than buttonsArea
self.controlArea.layout().addWidget(self.buttonsArea)
self.scene = QGraphicsScene()
self.view = QGraphicsView(self.scene)
self.view.setRenderHint(QPainter.Antialiasing, True)
self.view.setAlignment(Qt.AlignTop | Qt.AlignLeft)
self.mainArea.layout().addWidget(self.view)
def sizeHint(self):
sh = self.controlArea.sizeHint()
return sh.expandedTo(QSize(600, 720))
@check_sql_input
def set_data(self, data):
"""
Set the input data set.
"""
self.closeContext()
self.clear()
error_msg = ""
warning_msg = ""
candidatevars = []
if data is not None:
candidatevars = [
v for v in data.domain.variables + data.domain.metas
if v.is_discrete and len(v.values) >= 2]
if not candidatevars:
error_msg = "Input does not have any suitable cluster labels."
data = None
if data is not None:
ncont = sum(v.is_continuous for v in data.domain.attributes)
ndiscrete = len(data.domain.attributes) - ncont
if ncont == 0:
data = None
error_msg = "No continuous columns"
elif ncont < len(data.domain.attributes):
warning_msg = "{0} discrete columns will not be used for " \
"distance computation".format(ndiscrete)
self.data = data
if data is not None:
self.cluster_var_model[:] = candidatevars
if data.domain.class_var in candidatevars:
self.cluster_var_idx = \
candidatevars.index(data.domain.class_var)
else:
self.cluster_var_idx = 0
annotvars = [var for var in data.domain.metas if var.is_string]
self.annotation_var_model[:] = ["None"] + annotvars
self.annotation_var_idx = 1 if len(annotvars) else 0
self._effective_data = Orange.distance._preprocess(data)
self.openContext(Orange.data.Domain(candidatevars))
self.error(error_msg)
self.warning(warning_msg)
def handleNewSignals(self):
if self._effective_data is not None:
self._update()
self._replot()
self.unconditional_commit()
def clear(self):
"""
Clear the widget state.
"""
self.data = None
self._effective_data = None
self._matrix = None
self._silhouette = None
self._labels = None
self.cluster_var_model[:] = []
self.annotation_var_model[:] = ["None"]
self._clear_scene()
def _clear_scene(self):
# Clear the graphics scene and associated objects
self.scene.clear()
self.scene.setSceneRect(QRectF())
self._silplot = None
def _invalidate_distances(self):
# Invalidate the computed distance matrix and recompute the silhouette.
self._matrix = None
self._invalidate_scores()
def _invalidate_scores(self):
# Invalidate and recompute the current silhouette scores.
self._labels = self._silhouette = None
self._update()
self._replot()
if self.data is not None:
self.commit()
def _update(self):
# Update/recompute the distances/scores as required
if self.data is None:
self._silhouette = None
self._labels = None
self._matrix = None
self._clear_scene()
return
if self._matrix is None and self._effective_data is not None:
_, metric = self.Distances[self.distance_idx]
self._matrix = numpy.asarray(metric(self._effective_data))
labelvar = self.cluster_var_model[self.cluster_var_idx]
labels, _ = self.data.get_column_view(labelvar)
labels = labels.astype(int)
_, counts = numpy.unique(labels, return_counts=True)
if numpy.count_nonzero(counts) >= 2:
self.Error.need_two_clusters.clear()
silhouette = sklearn.metrics.silhouette_samples(
self._matrix, labels, metric="precomputed")
else:
self.Error.need_two_clusters()
labels = silhouette = None
self._labels = labels
self._silhouette = silhouette
def _set_bar_height(self):
visible = self.bar_size >= 5
self._silplot.setBarHeight(self.bar_size)
self._silplot.setRowNamesVisible(visible)
self.ann_hidden_warning.setVisible(
not visible and self.annotation_var_idx > 0)
def _replot(self):
# Clear and replot/initialize the scene
self._clear_scene()
if self._silhouette is not None and self._labels is not None:
var = self.cluster_var_model[self.cluster_var_idx]
self._silplot = silplot = SilhouettePlot()
self._set_bar_height()
if self.group_by_cluster:
silplot.setScores(self._silhouette, self._labels, var.values)
else:
silplot.setScores(
self._silhouette,
numpy.zeros(len(self._silhouette), dtype=int),
[""]
)
self.scene.addItem(silplot)
self._update_annotations()
silplot.resize(silplot.effectiveSizeHint(Qt.PreferredSize))
silplot.selectionChanged.connect(self.commit)
self.scene.setSceneRect(
QRectF(QPointF(0, 0),
self._silplot.effectiveSizeHint(Qt.PreferredSize)))
def _update_bar_size(self):
if self._silplot is not None:
self._set_bar_height()
self.scene.setSceneRect(
QRectF(QPointF(0, 0),
self._silplot.effectiveSizeHint(Qt.PreferredSize)))
def _update_annotations(self):
if 0 < self.annotation_var_idx < len(self.annotation_var_model):
annot_var = self.annotation_var_model[self.annotation_var_idx]
else:
annot_var = None
self.ann_hidden_warning.setVisible(
self.bar_size < 5 and annot_var is not None)
if self._silplot is not None:
if annot_var is not None:
column, _ = self.data.get_column_view(annot_var)
self._silplot.setRowNames(
[annot_var.str_val(value) for value in column])
else:
self._silplot.setRowNames(None)
def commit(self):
"""
Commit/send the current selection to the output.
"""
selected = indices = data = None
if self.data is not None:
selectedmask = numpy.full(len(self.data), False, dtype=bool)
if self._silplot is not None:
indices = self._silplot.selection()
selectedmask[indices] = True
scores = self._silhouette
silhouette_var = None
if self.add_scores:
var = self.cluster_var_model[self.cluster_var_idx]
silhouette_var = Orange.data.ContinuousVariable(
"Silhouette ({})".format(escape(var.name)))
domain = Orange.data.Domain(
self.data.domain.attributes,
self.data.domain.class_vars,
self.data.domain.metas + (silhouette_var, ))
data = self.data.from_table(
domain, self.data)
else:
domain = self.data.domain
data = self.data
if numpy.count_nonzero(selectedmask):
selected = self.data.from_table(
domain, self.data, numpy.flatnonzero(selectedmask))
if self.add_scores:
if selected is not None:
selected[:, silhouette_var] = numpy.c_[scores[selectedmask]]
data[:, silhouette_var] = numpy.c_[scores]
self.send("Selected Data", selected)
self.send(ANNOTATED_DATA_SIGNAL_NAME,
create_annotated_table(data, indices))
def send_report(self):
if not len(self.cluster_var_model):
return
self.report_plot()
caption = "Silhouette plot ({} distance), clustered by '{}'".format(
self.Distances[self.distance_idx][0],
self.cluster_var_model[self.cluster_var_idx])
if self.annotation_var_idx and self._silplot.rowNamesVisible():
caption += ", annotated with '{}'".format(
self.annotation_var_model[self.annotation_var_idx])
self.report_caption(caption)
def onDeleteWidget(self):
self.clear()
super().onDeleteWidget()
class OWSilhouettePlot(widget.OWWidget):
name = "Silhouette Plot"
description = "Visually assess cluster quality and " \
"the degree of cluster membership."
icon = "icons/SilhouettePlot.svg"
priority = 300
keywords = []
class Inputs:
data = Input("Data", Orange.data.Table)
class Outputs:
selected_data = Output("Selected Data",
Orange.data.Table,
default=True)
annotated_data = Output(ANNOTATED_DATA_SIGNAL_NAME, Orange.data.Table)
replaces = [
"orangecontrib.prototypes.widgets.owsilhouetteplot.OWSilhouettePlot",
"Orange.widgets.unsupervised.owsilhouetteplot.OWSilhouettePlot"
]
settingsHandler = settings.PerfectDomainContextHandler()
#: Distance metric index
distance_idx = settings.Setting(0)
#: Group/cluster variable index
cluster_var_idx = settings.ContextSetting(0)
#: Annotation variable index
annotation_var_idx = settings.ContextSetting(0)
#: Group the (displayed) silhouettes by cluster
group_by_cluster = settings.Setting(True)
#: A fixed size for an instance bar
bar_size = settings.Setting(3)
#: Add silhouette scores to output data
add_scores = settings.Setting(False)
auto_commit = settings.Setting(True)
Distances = [("Euclidean", Orange.distance.Euclidean),
("Manhattan", Orange.distance.Manhattan)]
graph_name = "scene"
buttons_area_orientation = Qt.Vertical
class Error(widget.OWWidget.Error):
need_two_clusters = Msg("Need at least two non-empty clusters")
singleton_clusters_all = Msg("All clusters are singletons")
memory_error = Msg("Not enough memory")
value_error = Msg("Distances could not be computed: '{}'")
class Warning(widget.OWWidget.Warning):
missing_cluster_assignment = Msg(
"{} instance{s} omitted (missing cluster assignment)")
def __init__(self):
super().__init__()
#: The input data
self.data = None # type: Optional[Orange.data.Table]
#: Distance matrix computed from data
self._matrix = None # type: Optional[Orange.misc.DistMatrix]
#: An bool mask (size == len(data)) indicating missing group/cluster
#: assignments
self._mask = None # type: Optional[np.ndarray]
#: An array of cluster/group labels for instances with valid group
#: assignment
self._labels = None # type: Optional[np.ndarray]
#: An array of silhouette scores for instances with valid group
#: assignment
self._silhouette = None # type: Optional[np.ndarray]
self._silplot = None # type: Optional[SilhouettePlot]
gui.comboBox(self.controlArea,
self,
"distance_idx",
box="Distance",
items=[name for name, _ in OWSilhouettePlot.Distances],
orientation=Qt.Horizontal,
callback=self._invalidate_distances)
box = gui.vBox(self.controlArea, "Cluster Label")
self.cluster_var_cb = gui.comboBox(box,
self,
"cluster_var_idx",
contentsLength=14,
addSpace=4,
callback=self._invalidate_scores)
gui.checkBox(box,
self,
"group_by_cluster",
"Group by cluster",
callback=self._replot)
self.cluster_var_model = itemmodels.VariableListModel(parent=self)
self.cluster_var_cb.setModel(self.cluster_var_model)
box = gui.vBox(self.controlArea, "Bars")
gui.widgetLabel(box, "Bar width:")
gui.hSlider(box,
self,
"bar_size",
minValue=1,
maxValue=10,
step=1,
callback=self._update_bar_size,
addSpace=6)
gui.widgetLabel(box, "Annotations:")
self.annotation_cb = gui.comboBox(box,
self,
"annotation_var_idx",
contentsLength=14,
callback=self._update_annotations)
self.annotation_var_model = itemmodels.VariableListModel(parent=self)
self.annotation_var_model[:] = ["None"]
self.annotation_cb.setModel(self.annotation_var_model)
ibox = gui.indentedBox(box, 5)
self.ann_hidden_warning = warning = gui.widgetLabel(
ibox, "(increase the width to show)")
ibox.setFixedWidth(ibox.sizeHint().width())
warning.setVisible(False)
gui.rubber(self.controlArea)
gui.separator(self.buttonsArea)
box = gui.vBox(self.buttonsArea, "Output")
# Thunk the call to commit to call conditional commit
gui.checkBox(box,
self,
"add_scores",
"Add silhouette scores",
callback=lambda: self.commit())
gui.auto_commit(box,
self,
"auto_commit",
"Commit",
auto_label="Auto commit",
box=False)
# Ensure that the controlArea is not narrower than buttonsArea
self.controlArea.layout().addWidget(self.buttonsArea)
self.scene = QGraphicsScene()
self.view = QGraphicsView(self.scene)
self.view.setRenderHint(QPainter.Antialiasing, True)
self.view.setAlignment(Qt.AlignTop | Qt.AlignLeft)
self.mainArea.layout().addWidget(self.view)
def sizeHint(self):
sh = self.controlArea.sizeHint()
return sh.expandedTo(QSize(600, 720))
@Inputs.data
@check_sql_input
def set_data(self, data):
"""
Set the input dataset.
"""
self.closeContext()
self.clear()
error_msg = ""
warning_msg = ""
candidatevars = []
if data is not None:
candidatevars = [
v for v in data.domain.variables + data.domain.metas
if v.is_discrete and len(v.values) >= 2
]
if not candidatevars:
error_msg = "Input does not have any suitable labels."
data = None
self.data = data
if data is not None:
self.cluster_var_model[:] = candidatevars
if data.domain.class_var in candidatevars:
self.cluster_var_idx = \
candidatevars.index(data.domain.class_var)
else:
self.cluster_var_idx = 0
annotvars = [var for var in data.domain.metas if var.is_string]
self.annotation_var_model[:] = ["None"] + annotvars
self.annotation_var_idx = 1 if len(annotvars) else 0
self.openContext(Orange.data.Domain(candidatevars))
self.error(error_msg)
self.warning(warning_msg)
def handleNewSignals(self):
if self.data is not None:
self._update()
self._replot()
self.unconditional_commit()
def clear(self):
"""
Clear the widget state.
"""
self.data = None
self._matrix = None
self._mask = None
self._silhouette = None
self._labels = None
self.cluster_var_model[:] = []
self.annotation_var_model[:] = ["None"]
self._clear_scene()
self.Error.clear()
self.Warning.clear()
def _clear_scene(self):
# Clear the graphics scene and associated objects
self.scene.clear()
self.scene.setSceneRect(QRectF())
self._silplot = None
def _invalidate_distances(self):
# Invalidate the computed distance matrix and recompute the silhouette.
self._matrix = None
self._invalidate_scores()
def _invalidate_scores(self):
# Invalidate and recompute the current silhouette scores.
self._labels = self._silhouette = self._mask = None
self._update()
self._replot()
if self.data is not None:
self.commit()
def _update(self):
# Update/recompute the distances/scores as required
self._clear_messages()
if self.data is None or not len(self.data):
self._reset_all()
return
if self._matrix is None and self.data is not None:
_, metric = self.Distances[self.distance_idx]
try:
self._matrix = np.asarray(metric(self.data))
except MemoryError:
self.Error.memory_error()
return
except ValueError as err:
self.Error.value_error(str(err))
return
self._update_labels()
def _reset_all(self):
self._mask = None
self._silhouette = None
self._labels = None
self._matrix = None
self._clear_scene()
def _clear_messages(self):
self.Error.clear()
self.Warning.missing_cluster_assignment.clear()
def _update_labels(self):
labelvar = self.cluster_var_model[self.cluster_var_idx]
labels, _ = self.data.get_column_view(labelvar)
labels = np.asarray(labels, dtype=float)
mask = np.isnan(labels)
labels = labels.astype(int)
labels = labels[~mask]
labels_unq, _ = np.unique(labels, return_counts=True)
if len(labels_unq) < 2:
self.Error.need_two_clusters()
labels = silhouette = mask = None
elif len(labels_unq) == len(labels):
self.Error.singleton_clusters_all()
labels = silhouette = mask = None
else:
silhouette = sklearn.metrics.silhouette_samples(
self._matrix[~mask, :][:, ~mask], labels, metric="precomputed")
self._mask = mask
self._labels = labels
self._silhouette = silhouette
if labels is not None:
count_missing = np.count_nonzero(mask)
if count_missing:
self.Warning.missing_cluster_assignment(
count_missing, s="s" if count_missing > 1 else "")
def _set_bar_height(self):
visible = self.bar_size >= 5
self._silplot.setBarHeight(self.bar_size)
self._silplot.setRowNamesVisible(visible)
self.ann_hidden_warning.setVisible(not visible
and self.annotation_var_idx > 0)
def _replot(self):
# Clear and replot/initialize the scene
self._clear_scene()
if self._silhouette is not None and self._labels is not None:
var = self.cluster_var_model[self.cluster_var_idx]
self._silplot = silplot = SilhouettePlot()
self._set_bar_height()
if self.group_by_cluster:
silplot.setScores(self._silhouette, self._labels, var.values,
var.colors)
else:
silplot.setScores(self._silhouette,
np.zeros(len(self._silhouette), dtype=int),
[""], np.array([[63, 207, 207]]))
self.scene.addItem(silplot)
self._update_annotations()
silplot.selectionChanged.connect(self.commit)
silplot.layout().activate()
self._update_scene_rect()
silplot.geometryChanged.connect(self._update_scene_rect)
def _update_bar_size(self):
if self._silplot is not None:
self._set_bar_height()
def _update_annotations(self):
if 0 < self.annotation_var_idx < len(self.annotation_var_model):
annot_var = self.annotation_var_model[self.annotation_var_idx]
else:
annot_var = None
self.ann_hidden_warning.setVisible(self.bar_size < 5
and annot_var is not None)
if self._silplot is not None:
if annot_var is not None:
column, _ = self.data.get_column_view(annot_var)
if self._mask is not None:
assert column.shape == self._mask.shape
column = column[~self._mask]
self._silplot.setRowNames(
[annot_var.str_val(value) for value in column])
else:
self._silplot.setRowNames(None)
def _update_scene_rect(self):
self.scene.setSceneRect(self._silplot.geometry())
def commit(self):
"""
Commit/send the current selection to the output.
"""
selected = indices = data = None
if self.data is not None:
selectedmask = np.full(len(self.data), False, dtype=bool)
if self._silplot is not None:
indices = self._silplot.selection()
assert (np.diff(indices) > 0).all(), "strictly increasing"
if self._mask is not None:
indices = np.flatnonzero(~self._mask)[indices]
selectedmask[indices] = True
if self._mask is not None:
scores = np.full(shape=selectedmask.shape, fill_value=np.nan)
scores[~self._mask] = self._silhouette
else:
scores = self._silhouette
silhouette_var = None
if self.add_scores:
var = self.cluster_var_model[self.cluster_var_idx]
silhouette_var = Orange.data.ContinuousVariable(
"Silhouette ({})".format(escape(var.name)))
domain = Orange.data.Domain(
self.data.domain.attributes, self.data.domain.class_vars,
self.data.domain.metas + (silhouette_var, ))
data = self.data.transform(domain)
else:
domain = self.data.domain
data = self.data
if np.count_nonzero(selectedmask):
selected = self.data.from_table(domain, self.data,
np.flatnonzero(selectedmask))
if self.add_scores:
if selected is not None:
selected[:, silhouette_var] = np.c_[scores[selectedmask]]
data[:, silhouette_var] = np.c_[scores]
self.Outputs.selected_data.send(selected)
self.Outputs.annotated_data.send(create_annotated_table(data, indices))
def send_report(self):
if not len(self.cluster_var_model):
return
self.report_plot()
caption = "Silhouette plot ({} distance), clustered by '{}'".format(
self.Distances[self.distance_idx][0],
self.cluster_var_model[self.cluster_var_idx])
if self.annotation_var_idx and self._silplot.rowNamesVisible():
caption += ", annotated with '{}'".format(
self.annotation_var_model[self.annotation_var_idx])
self.report_caption(caption)
def onDeleteWidget(self):
self.clear()
super().onDeleteWidget()
class OWMosaicDisplay(OWWidget):
name = "Mosaic Display"
description = "Display data in a mosaic plot."
icon = "icons/MosaicDisplay.svg"
priority = 220
keywords = []
class Inputs:
data = Input("Data", Table, default=True)
data_subset = Input("Data Subset", Table)
class Outputs:
selected_data = Output("Selected Data", Table, default=True)
annotated_data = Output(ANNOTATED_DATA_SIGNAL_NAME, Table)
settingsHandler = DomainContextHandler()
vizrank = SettingProvider(MosaicVizRank)
settings_version = 2
use_boxes = Setting(True)
variable1: Variable = ContextSetting(None)
variable2: Variable = ContextSetting(None)
variable3: Variable = ContextSetting(None)
variable4: Variable = ContextSetting(None)
variable_color: DiscreteVariable = ContextSetting(None)
selection = Setting(set(), schema_only=True)
BAR_WIDTH = 5
SPACING = 4
ATTR_NAME_OFFSET = 20
ATTR_VAL_OFFSET = 3
BLUE_COLORS = [
QColor(255, 255, 255),
QColor(210, 210, 255),
QColor(110, 110, 255),
QColor(0, 0, 255)
]
RED_COLORS = [
QColor(255, 255, 255),
QColor(255, 200, 200),
QColor(255, 100, 100),
QColor(255, 0, 0)
]
graph_name = "canvas"
attrs_changed_manually = Signal(list)
class Warning(OWWidget.Warning):
incompatible_subset = Msg("Data subset is incompatible with Data")
no_valid_data = Msg("No valid data")
no_cont_selection_sql = \
Msg("Selection of numeric features on SQL is not supported")
def __init__(self):
super().__init__()
self.data = None
self.discrete_data = None
self.subset_data = None
self.subset_indices = None
self.__pending_selection = self.selection
self.selection = set()
self.color_data = None
self.areas = []
self.canvas = QGraphicsScene(self)
self.canvas_view = ViewWithPress(self.canvas,
handler=self.clear_selection)
self.mainArea.layout().addWidget(self.canvas_view)
self.canvas_view.setVerticalScrollBarPolicy(Qt.ScrollBarAlwaysOff)
self.canvas_view.setHorizontalScrollBarPolicy(Qt.ScrollBarAlwaysOff)
self.canvas_view.setRenderHint(QPainter.Antialiasing)
box = gui.vBox(self.controlArea, box=True)
self.model_1 = DomainModel(order=DomainModel.MIXED,
valid_types=DomainModel.PRIMITIVE)
self.model_234 = DomainModel(order=DomainModel.MIXED,
valid_types=DomainModel.PRIMITIVE,
placeholder="(None)")
self.attr_combos = [
gui.comboBox(box,
self,
value="variable{}".format(i),
orientation=Qt.Horizontal,
contentsLength=12,
searchable=True,
callback=self.attr_changed,
model=self.model_1 if i == 1 else self.model_234)
for i in range(1, 5)
]
self.vizrank, self.vizrank_button = MosaicVizRank.add_vizrank(
box, self, "Find Informative Mosaics", self.set_attr)
box2 = gui.vBox(self.controlArea, box="Interior Coloring")
self.color_model = DomainModel(order=DomainModel.MIXED,
valid_types=DomainModel.PRIMITIVE,
placeholder="(Pearson residuals)")
self.cb_attr_color = gui.comboBox(box2,
self,
value="variable_color",
orientation=Qt.Horizontal,
contentsLength=12,
labelWidth=50,
searchable=True,
callback=self.set_color_data,
model=self.color_model)
self.bar_button = gui.checkBox(box2,
self,
'use_boxes',
label='Compare with total',
callback=self.update_graph)
gui.rubber(self.controlArea)
def sizeHint(self):
return QSize(720, 530)
def _get_discrete_data(self, data):
"""
Discretize continuous attributes.
Return None when there is no data, no rows, or no primitive attributes.
"""
if (data is None or not len(data) or not any(
attr.is_discrete or attr.is_continuous
for attr in chain(data.domain.variables, data.domain.metas))):
return None
elif any(attr.is_continuous for attr in data.domain.variables):
return Discretize(method=EqualFreq(n=4),
remove_const=False,
discretize_classes=True,
discretize_metas=True)(data)
else:
return data
def init_combos(self, data):
def set_combos(value):
self.model_1.set_domain(value)
self.model_234.set_domain(value)
self.color_model.set_domain(value)
if data is None:
set_combos(None)
self.variable1 = self.variable2 = self.variable3 \
= self.variable4 = self.variable_color = None
return
set_combos(self.data.domain)
if len(self.model_1) > 0:
self.variable1 = self.model_1[0]
self.variable2 = self.model_1[min(1, len(self.model_1) - 1)]
self.variable3 = self.variable4 = None
self.variable_color = self.data.domain.class_var # None is OK, too
def get_disc_attr_list(self):
return [
self.discrete_data.domain[var.name]
for var in (self.variable1, self.variable2, self.variable3,
self.variable4) if var
]
def set_attr(self, *attrs):
self.variable1, self.variable2, self.variable3, self.variable4 = [
attr and self.data.domain[attr.name] for attr in attrs
]
self.reset_graph()
def attr_changed(self):
self.attrs_changed_manually.emit(self.get_disc_attr_list())
self.reset_graph()
def resizeEvent(self, e):
OWWidget.resizeEvent(self, e)
self.update_graph()
def showEvent(self, ev):
OWWidget.showEvent(self, ev)
self.update_graph()
@Inputs.data
def set_data(self, data):
if isinstance(data, SqlTable) and data.approx_len() > LARGE_TABLE:
data = data.sample_time(DEFAULT_SAMPLE_TIME)
self.closeContext()
self.data = data
self.vizrank.stop_and_reset()
self.vizrank_button.setEnabled(
self.data is not None and len(self.data) > 1
and len(self.data.domain.attributes) >= 1)
if self.data is None:
self.discrete_data = None
self.init_combos(None)
return
self.init_combos(self.data)
self.openContext(self.data)
@Inputs.data_subset
def set_subset_data(self, data):
self.subset_data = data
# this is called by widget after setData and setSubsetData are called.
# this way the graph is updated only once
def handleNewSignals(self):
self.Warning.incompatible_subset.clear()
self.subset_indices = None
if self.data is not None and self.subset_data:
transformed = self.subset_data.transform(self.data.domain)
if np.all(np.isnan(transformed.X)) \
and np.all(np.isnan(transformed.Y)):
self.Warning.incompatible_subset()
else:
indices = {e.id for e in transformed}
self.subset_indices = [ex.id in indices for ex in self.data]
if self.data is not None and self.__pending_selection is not None:
self.selection = self.__pending_selection
self.__pending_selection = None
else:
self.selection = set()
self.set_color_data()
self.update_graph()
self.send_selection()
def clear_selection(self):
self.selection = set()
self.update_selection_rects()
self.send_selection()
def coloring_changed(self):
self.vizrank.coloring_changed()
self.update_graph()
def reset_graph(self):
self.clear_selection()
self.update_graph()
def set_color_data(self):
if self.data is None:
return
self.bar_button.setEnabled(self.variable_color is not None)
attrs = [v for v in self.model_1 if v and v is not self.variable_color]
domain = Domain(attrs, self.variable_color, None)
self.color_data = self.data.from_table(domain, self.data)
self.discrete_data = self._get_discrete_data(self.color_data)
self.vizrank.stop_and_reset()
self.vizrank_button.setEnabled(True)
self.coloring_changed()
def update_selection_rects(self):
pens = (QPen(), QPen(Qt.black, 3, Qt.DotLine))
for i, (_, _, area) in enumerate(self.areas):
area.setPen(pens[i in self.selection])
def select_area(self, index, ev):
if ev.button() != Qt.LeftButton:
return
if ev.modifiers() & Qt.ControlModifier:
self.selection ^= {index}
else:
self.selection = {index}
self.update_selection_rects()
self.send_selection()
def send_selection(self):
if not self.selection or self.data is None:
self.Outputs.selected_data.send(None)
self.Outputs.annotated_data.send(
create_annotated_table(self.data, []))
return
filters = []
self.Warning.no_cont_selection_sql.clear()
if self.discrete_data is not self.data:
if isinstance(self.data, SqlTable):
self.Warning.no_cont_selection_sql()
for i in self.selection:
cols, vals, _ = self.areas[i]
filters.append(
filter.Values(
filter.FilterDiscrete(col, [val])
for col, val in zip(cols, vals)))
if len(filters) > 1:
filters = filter.Values(filters, conjunction=False)
else:
filters = filters[0]
selection = filters(self.discrete_data)
idset = set(selection.ids)
sel_idx = [i for i, id in enumerate(self.data.ids) if id in idset]
if self.discrete_data is not self.data:
selection = self.data[sel_idx]
self.Outputs.selected_data.send(selection)
self.Outputs.annotated_data.send(
create_annotated_table(self.data, sel_idx))
def send_report(self):
self.report_plot(self.canvas)
def update_graph(self):
spacing = self.SPACING
bar_width = self.BAR_WIDTH
def get_counts(attr_vals, values):
"""Calculate rectangles' widths; if all are 0, they are set to 1."""
if not attr_vals:
counts = [conditionaldict[val] for val in values]
else:
counts = [
conditionaldict[attr_vals + "-" + val] for val in values
]
total = sum(counts)
if total == 0:
counts = [1] * len(values)
total = sum(counts)
return total, counts
def draw_data(attr_list,
x0_x1,
y0_y1,
side,
condition,
total_attrs,
used_attrs,
used_vals,
attr_vals=""):
x0, x1 = x0_x1
y0, y1 = y0_y1
if conditionaldict[attr_vals] == 0:
add_rect(x0,
x1,
y0,
y1,
"",
used_attrs,
used_vals,
attr_vals=attr_vals)
# store coordinates for later drawing of labels
draw_text(side, attr_list[0], (x0, x1), (y0, y1), total_attrs,
used_attrs, used_vals, attr_vals)
return
attr = attr_list[0]
# how much smaller rectangles do we draw
edge = len(attr_list) * spacing
values = get_variable_values_sorted(attr)
if side % 2:
values = values[::-1] # reverse names if necessary
if side % 2 == 0: # we are drawing on the x axis
# remove the space needed for separating different attr. values
whole = max(0, (x1 - x0) - edge * (len(values) - 1))
if whole == 0:
edge = (x1 - x0) / float(len(values) - 1)
else: # we are drawing on the y axis
whole = max(0, (y1 - y0) - edge * (len(values) - 1))
if whole == 0:
edge = (y1 - y0) / float(len(values) - 1)
total, counts = get_counts(attr_vals, values)
# when visualizing the third attribute and the first attribute has
# the last value, reverse the order in which the boxes are drawn;
# otherwise, if the last cell, nearest to the labels of the fourth
# attribute, is empty, we wouldn't be able to position the labels
valrange = list(range(len(values)))
if len(attr_list + used_attrs) == 4 and len(used_attrs) == 2:
attr1values = get_variable_values_sorted(used_attrs[0])
if used_vals[0] == attr1values[-1]:
valrange = valrange[::-1]
for i in valrange:
start = i * edge + whole * float(sum(counts[:i]) / total)
end = i * edge + whole * float(sum(counts[:i + 1]) / total)
val = values[i]
htmlval = to_html(val)
newattrvals = attr_vals + "-" + val if attr_vals else val
tooltip = "{} {}: <b>{}</b><br/>".format(
condition, attr.name, htmlval)
attrs = used_attrs + [attr]
vals = used_vals + [val]
args = attrs, vals, newattrvals
if side % 2 == 0: # if we are moving horizontally
if len(attr_list) == 1:
add_rect(x0 + start, x0 + end, y0, y1, tooltip, *args)
else:
draw_data(attr_list[1:], (x0 + start, x0 + end),
(y0, y1), side + 1, tooltip, total_attrs,
*args)
else:
if len(attr_list) == 1:
add_rect(x0, x1, y0 + start, y0 + end, tooltip, *args)
else:
draw_data(attr_list[1:], (x0, x1),
(y0 + start, y0 + end), side + 1, tooltip,
total_attrs, *args)
draw_text(side, attr_list[0], (x0, x1), (y0, y1), total_attrs,
used_attrs, used_vals, attr_vals)
def draw_text(side, attr, x0_x1, y0_y1, total_attrs, used_attrs,
used_vals, attr_vals):
x0, x1 = x0_x1
y0, y1 = y0_y1
if side in drawn_sides:
return
# the text on the right will be drawn when we are processing
# visualization of the last value of the first attribute
if side == 3:
attr1values = get_variable_values_sorted(used_attrs[0])
if used_vals[0] != attr1values[-1]:
return
if not conditionaldict[attr_vals]:
if side not in draw_positions:
draw_positions[side] = (x0, x1, y0, y1)
return
else:
if side in draw_positions:
# restore the positions of attribute values and name
(x0, x1, y0, y1) = draw_positions[side]
drawn_sides.add(side)
values = get_variable_values_sorted(attr)
if side % 2:
values = values[::-1]
spaces = spacing * (total_attrs - side) * (len(values) - 1)
width = x1 - x0 - spaces * (side % 2 == 0)
height = y1 - y0 - spaces * (side % 2 == 1)
# calculate position of first attribute
currpos = 0
total, counts = get_counts(attr_vals, values)
aligns = [
Qt.AlignTop | Qt.AlignHCenter, Qt.AlignRight | Qt.AlignVCenter,
Qt.AlignBottom | Qt.AlignHCenter,
Qt.AlignLeft | Qt.AlignVCenter
]
align = aligns[side]
for i, val in enumerate(values):
if distributiondict[val] != 0:
perc = counts[i] / float(total)
rwidth = width * perc
xs = [
x0 + currpos + rwidth / 2, x0 - self.ATTR_VAL_OFFSET,
x0 + currpos + rwidth / 2, x1 + self.ATTR_VAL_OFFSET
]
ys = [
y1 + self.ATTR_VAL_OFFSET,
y0 + currpos + height * 0.5 * perc,
y0 - self.ATTR_VAL_OFFSET,
y0 + currpos + height * 0.5 * perc
]
CanvasText(self.canvas,
val,
xs[side],
ys[side],
align,
max_width=rwidth if side == 0 else None)
space = height if side % 2 else width
currpos += perc * space + spacing * (total_attrs - side)
xs = [
x0 + (x1 - x0) / 2, x0 - max_ylabel_w1 - self.ATTR_VAL_OFFSET,
x0 + (x1 - x0) / 2, x1 + max_ylabel_w2 + self.ATTR_VAL_OFFSET
]
ys = [
y1 + self.ATTR_VAL_OFFSET + self.ATTR_NAME_OFFSET,
y0 + (y1 - y0) / 2,
y0 - self.ATTR_VAL_OFFSET - self.ATTR_NAME_OFFSET,
y0 + (y1 - y0) / 2
]
CanvasText(self.canvas,
attr.name,
xs[side],
ys[side],
align,
bold=True,
vertical=side % 2)
def add_rect(x0,
x1,
y0,
y1,
condition,
used_attrs,
used_vals,
attr_vals=""):
area_index = len(self.areas)
x1 += (x0 == x1)
y1 += (y0 == y1)
# rectangles of width and height 1 are not shown - increase
y1 += (x1 - x0 + y1 - y0 == 2)
colors = class_var and [QColor(*col) for col in class_var.colors]
def select_area(_, ev):
self.select_area(area_index, ev)
def rect(x, y, w, h, z, pen_color=None, brush_color=None, **args):
if pen_color is None:
return CanvasRectangle(self.canvas,
x,
y,
w,
h,
z=z,
onclick=select_area,
**args)
if brush_color is None:
brush_color = pen_color
return CanvasRectangle(self.canvas,
x,
y,
w,
h,
pen_color,
brush_color,
z=z,
onclick=select_area,
**args)
def line(x1, y1, x2, y2):
r = QGraphicsLineItem(x1, y1, x2, y2, None)
self.canvas.addItem(r)
r.setPen(QPen(Qt.white, 2))
r.setZValue(30)
outer_rect = rect(x0, y0, x1 - x0, y1 - y0, 30)
self.areas.append((used_attrs, used_vals, outer_rect))
if not conditionaldict[attr_vals]:
return
if self.variable_color is None:
s = sum(apriori_dists[0])
expected = s * reduce(
mul, (apriori_dists[i][used_vals[i]] / float(s)
for i in range(len(used_vals))))
actual = conditionaldict[attr_vals]
pearson = float((actual - expected) / sqrt(expected))
if pearson == 0:
ind = 0
else:
ind = max(0, min(int(log(abs(pearson), 2)), 3))
color = [self.RED_COLORS, self.BLUE_COLORS][pearson > 0][ind]
rect(x0, y0, x1 - x0, y1 - y0, -20, color)
outer_rect.setToolTip(
condition + "<hr/>" + "Expected instances: %.1f<br>"
"Actual instances: %d<br>"
"Standardized (Pearson) residual: %.1f" %
(expected, conditionaldict[attr_vals], pearson))
else:
cls_values = get_variable_values_sorted(class_var)
prior = get_distribution(data, class_var.name)
total = 0
for i, value in enumerate(cls_values):
val = conditionaldict[attr_vals + "-" + value]
if val == 0:
continue
if i == len(cls_values) - 1:
v = y1 - y0 - total
else:
v = ((y1 - y0) * val) / conditionaldict[attr_vals]
rect(x0, y0 + total, x1 - x0, v, -20, colors[i])
total += v
if self.use_boxes and \
abs(x1 - x0) > bar_width and abs(y1 - y0) > bar_width:
total = 0
line(x0 + bar_width, y0, x0 + bar_width, y1)
n = sum(prior)
for i, (val, color) in enumerate(zip(prior, colors)):
if i == len(prior) - 1:
h = y1 - y0 - total
else:
h = (y1 - y0) * val / n
rect(x0, y0 + total, bar_width, h, 20, color)
total += h
if conditionalsubsetdict:
if conditionalsubsetdict[attr_vals]:
if self.subset_indices is not None:
line(x1 - bar_width, y0, x1 - bar_width, y1)
total = 0
n = conditionalsubsetdict[attr_vals]
if n:
for i, (cls, color) in \
enumerate(zip(cls_values, colors)):
val = conditionalsubsetdict[attr_vals +
"-" + cls]
if val == 0:
continue
if i == len(prior) - 1:
v = y1 - y0 - total
else:
v = ((y1 - y0) * val) / n
rect(x1 - bar_width, y0 + total, bar_width,
v, 15, color)
total += v
actual = [
conditionaldict[attr_vals + "-" + cls_values[i]]
for i in range(len(prior))
]
n_actual = sum(actual)
if n_actual > 0:
apriori = [prior[key] for key in cls_values]
n_apriori = sum(apriori)
text = "<br/>".join(
"<b>%s</b>: %d / %.1f%% (Expected %.1f / %.1f%%)" %
(cls, act, 100.0 * act / n_actual,
apr / n_apriori * n_actual, 100.0 * apr / n_apriori)
for cls, act, apr in zip(cls_values, actual, apriori))
else:
text = ""
outer_rect.setToolTip("{}<hr>Instances: {}<br><br>{}".format(
condition, n_actual, text[:-4]))
def create_legend():
if self.variable_color is None:
names = [
"<-8", "-8:-4", "-4:-2", "-2:2", "2:4", "4:8", ">8",
"Residuals:"
]
colors = self.RED_COLORS[::-1] + self.BLUE_COLORS[1:]
edges = repeat(Qt.black)
else:
names = get_variable_values_sorted(class_var)
edges = colors = [QColor(*col) for col in class_var.colors]
items = []
size = 8
for name, color, edgecolor in zip(names, colors, edges):
item = QGraphicsItemGroup()
item.addToGroup(
CanvasRectangle(None, -size / 2, -size / 2, size, size,
edgecolor, color))
item.addToGroup(
CanvasText(None, name, size, 0, Qt.AlignVCenter))
items.append(item)
return wrap_legend_items(items,
hspacing=20,
vspacing=16 + size,
max_width=self.canvas_view.width() - xoff)
self.canvas.clear()
self.areas = []
data = self.discrete_data
if data is None:
return
attr_list = self.get_disc_attr_list()
class_var = data.domain.class_var
# TODO: check this
# data = Preprocessor_dropMissing(data)
unique = [v.name for v in set(attr_list + [class_var]) if v]
if len(data[:, unique]) == 0:
self.Warning.no_valid_data()
return
else:
self.Warning.no_valid_data.clear()
attrs = [attr for attr in attr_list if not attr.values]
if attrs:
CanvasText(self.canvas,
"Feature {} has no values".format(attrs[0]),
(self.canvas_view.width() - 120) / 2,
self.canvas_view.height() / 2)
return
if self.variable_color is None:
apriori_dists = [
get_distribution(data, attr) for attr in attr_list
]
else:
apriori_dists = []
def get_max_label_width(attr):
values = get_variable_values_sorted(attr)
maxw = 0
for val in values:
t = CanvasText(self.canvas, val, 0, 0, bold=0, show=False)
maxw = max(int(t.boundingRect().width()), maxw)
return maxw
xoff = 20
# get the maximum width of rectangle
width = 20
max_ylabel_w1 = max_ylabel_w2 = 0
if len(attr_list) > 1:
text = CanvasText(self.canvas, attr_list[1].name, bold=1, show=0)
max_ylabel_w1 = min(get_max_label_width(attr_list[1]), 150)
width = 5 + text.boundingRect().height() + \
self.ATTR_VAL_OFFSET + max_ylabel_w1
xoff = width
if len(attr_list) == 4:
text = CanvasText(self.canvas,
attr_list[3].name,
bold=1,
show=0)
max_ylabel_w2 = min(get_max_label_width(attr_list[3]), 150)
width += text.boundingRect().height() + \
self.ATTR_VAL_OFFSET + max_ylabel_w2 - 10
legend = create_legend()
# get the maximum height of rectangle
yoff = 45
legendoff = yoff + self.ATTR_NAME_OFFSET + self.ATTR_VAL_OFFSET + 35
square_size = min(
self.canvas_view.width() - width - 20,
self.canvas_view.height() - legendoff -
legend.boundingRect().height())
if square_size < 0:
return # canvas is too small to draw rectangles
self.canvas_view.setSceneRect(0, 0, self.canvas_view.width(),
self.canvas_view.height())
drawn_sides = set()
draw_positions = {}
conditionaldict, distributiondict = \
get_conditional_distribution(data, attr_list)
conditionalsubsetdict = None
if self.subset_indices:
conditionalsubsetdict, _ = get_conditional_distribution(
self.discrete_data[self.subset_indices], attr_list)
# draw rectangles
draw_data(attr_list, (xoff, xoff + square_size),
(yoff, yoff + square_size), 0, "", len(attr_list), [], [])
self.canvas.addItem(legend)
legend.setPos(
xoff - legend.boundingRect().x() +
max(0, (square_size - legend.boundingRect().width()) / 2),
legendoff + square_size)
self.update_selection_rects()
@classmethod
def migrate_context(cls, context, version):
if version < 2:
settings.migrate_str_to_variable(context,
none_placeholder="(None)")
class OWBoxPlot(widget.OWWidget):
"""
Here's how the widget's functions call each other:
- `set_data` is a signal handler fills the list boxes and calls
`grouping_changed`.
- `grouping_changed` handles changes of grouping attribute: it enables or
disables the box for ordering, orders attributes and calls `attr_changed`.
- `attr_changed` handles changes of attribute. It recomputes box data by
calling `compute_box_data`, shows the appropriate display box
(discrete/continuous) and then calls`layout_changed`
- `layout_changed` constructs all the elements for the scene (as lists of
QGraphicsItemGroup) and calls `display_changed`. It is called when the
attribute or grouping is changed (by attr_changed) and on resize event.
- `display_changed` puts the elements corresponding to the current display
settings on the scene. It is called when the elements are reconstructed
(layout is changed due to selection of attributes or resize event), or
when the user changes display settings or colors.
For discrete attributes, the flow is a bit simpler: the elements are not
constructed in advance (by layout_changed). Instead, layout_changed and
display_changed call display_changed_disc that draws everything.
"""
name = "箱形图"
description = "在方框图中可视化特征值的分布。"
icon = "icons/BoxPlot.svg"
priority = 100
keywords = ["whisker"]
class Inputs:
data = Input("数据", Orange.data.Table)
class Outputs:
selected_data = Output("所选数据", Orange.data.Table, default=True)
annotated_data = Output(ANNOTATED_DATA_SIGNAL_NAME, Orange.data.Table)
#: Comparison types for continuous variables
CompareNone, CompareMedians, CompareMeans = 0, 1, 2
settingsHandler = DomainContextHandler()
conditions = ContextSetting([])
attribute = ContextSetting(None)
order_by_importance = Setting(False)
group_var = ContextSetting(None)
show_annotations = Setting(True)
compare = Setting(CompareMeans)
stattest = Setting(0)
sig_threshold = Setting(0.05)
stretched = Setting(True)
show_labels = Setting(True)
sort_freqs = Setting(False)
auto_commit = Setting(True)
_sorting_criteria_attrs = {
CompareNone: "",
CompareMedians: "median",
CompareMeans: "mean"
}
_pen_axis_tick = QPen(Qt.white, 5)
_pen_axis = QPen(Qt.darkGray, 3)
_pen_median = QPen(QBrush(QColor(0xff, 0xff, 0x00)), 2)
_pen_paramet = QPen(QBrush(QColor(0x33, 0x00, 0xff)), 2)
_pen_dotted = QPen(QBrush(QColor(0x33, 0x00, 0xff)), 1)
_pen_dotted.setStyle(Qt.DotLine)
_post_line_pen = QPen(Qt.lightGray, 2)
_post_grp_pen = QPen(Qt.lightGray, 4)
for pen in (_pen_paramet, _pen_median, _pen_dotted, _pen_axis,
_pen_axis_tick, _post_line_pen, _post_grp_pen):
pen.setCosmetic(True)
pen.setCapStyle(Qt.RoundCap)
pen.setJoinStyle(Qt.RoundJoin)
_pen_axis_tick.setCapStyle(Qt.FlatCap)
_box_brush = QBrush(QColor(0x33, 0x88, 0xff, 0xc0))
_axis_font = QFont()
_axis_font.setPixelSize(12)
_label_font = QFont()
_label_font.setPixelSize(11)
_attr_brush = QBrush(QColor(0x33, 0x00, 0xff))
graph_name = "盒式布景"
def __init__(self):
super().__init__()
self.stats = []
self.dataset = None
self.posthoc_lines = []
self.label_txts = self.mean_labels = self.boxes = self.labels = \
self.label_txts_all = self.attr_labels = self.order = []
self.p = -1.0
self.scale_x = self.scene_min_x = self.scene_width = 0
self.label_width = 0
self.attrs = VariableListModel()
view = gui.listView(self.controlArea,
self,
"attribute",
box="变量",
model=self.attrs,
callback=self.attr_changed)
view.setMinimumSize(QSize(30, 30))
# Any other policy than Ignored will let the QListBox's scrollbar
# set the minimal height (see the penultimate paragraph of
# http://doc.qt.io/qt-4.8/qabstractscrollarea.html#addScrollBarWidget)
view.setSizePolicy(QSizePolicy.Expanding, QSizePolicy.Ignored)
gui.separator(view.box, 6, 6)
self.cb_order = gui.checkBox(view.box,
self,
"order_by_importance",
"按相关性排序",
tooltip="由𝜒²或方差对子群排序",
callback=self.apply_sorting)
self.group_vars = DomainModel(placeholder="无",
separators=False,
valid_types=Orange.data.DiscreteVariable)
self.group_view = view = gui.listView(self.controlArea,
self,
"group_var",
box="子群",
model=self.group_vars,
callback=self.grouping_changed)
view.setEnabled(False)
view.setMinimumSize(QSize(30, 30))
# See the comment above
view.setSizePolicy(QSizePolicy.Expanding, QSizePolicy.Ignored)
# TODO: move Compare median/mean to grouping box
# The vertical size policy is needed to let only the list views expand
self.display_box = gui.vBox(self.controlArea,
"Display",
sizePolicy=(QSizePolicy.Minimum,
QSizePolicy.Maximum),
addSpace=False)
gui.checkBox(self.display_box,
self,
"show_annotations",
"Annotate",
callback=self.display_changed)
self.compare_rb = gui.radioButtonsInBox(
self.display_box,
self,
'compare',
btnLabels=["无比较", "中位数比较", "均值比较"],
callback=self.layout_changed)
# The vertical size policy is needed to let only the list views expand
self.stretching_box = box = gui.vBox(self.controlArea,
box="显示",
sizePolicy=(QSizePolicy.Minimum,
QSizePolicy.Fixed))
self.stretching_box.sizeHint = self.display_box.sizeHint
gui.checkBox(box,
self,
'stretched',
"拉杆",
callback=self.display_changed)
gui.checkBox(box,
self,
'show_labels',
"显示框标签",
callback=self.display_changed)
self.sort_cb = gui.checkBox(box,
self,
'sort_freqs',
"按子组频率排序",
callback=self.display_changed)
gui.rubber(box)
gui.auto_commit(self.controlArea, self, "auto_commit", "选中发送", "自动发送")
gui.vBox(self.mainArea, addSpace=True)
self.box_scene = QGraphicsScene()
self.box_scene.selectionChanged.connect(self.commit)
self.box_view = QGraphicsView(self.box_scene)
self.box_view.setRenderHints(QPainter.Antialiasing
| QPainter.TextAntialiasing
| QPainter.SmoothPixmapTransform)
self.box_view.viewport().installEventFilter(self)
self.mainArea.layout().addWidget(self.box_view)
e = gui.hBox(self.mainArea, addSpace=False)
self.infot1 = gui.widgetLabel(e, "<center>没有测试结果。</center>")
self.mainArea.setMinimumWidth(600)
self.stats = self.dist = self.conts = []
self.is_continuous = False
self.update_display_box()
def sizeHint(self):
return QSize(100, 500) # Vertical size is regulated by mainArea
def eventFilter(self, obj, event):
if obj is self.box_view.viewport() and \
event.type() == QEvent.Resize:
self.layout_changed()
return super().eventFilter(obj, event)
def reset_attrs(self, domain):
self.attrs[:] = [
var for var in chain(domain.class_vars, domain.metas,
domain.attributes) if var.is_primitive()
]
# noinspection PyTypeChecker
@Inputs.data
def set_data(self, dataset):
if dataset is not None and (not bool(dataset)
or not len(dataset.domain) and
not any(var.is_primitive()
for var in dataset.domain.metas)):
dataset = None
self.closeContext()
self.dataset = dataset
self.dist = self.stats = self.conts = []
self.group_var = None
self.attribute = None
if dataset:
domain = dataset.domain
self.group_vars.set_domain(domain)
self.group_view.setEnabled(len(self.group_vars) > 1)
self.reset_attrs(domain)
self.select_default_variables(domain)
self.openContext(self.dataset)
self.grouping_changed()
else:
self.reset_all_data()
self.commit()
def select_default_variables(self, domain):
# visualize first non-class variable, group by class (if present)
if len(self.attrs) > len(domain.class_vars):
self.attribute = self.attrs[len(domain.class_vars)]
elif self.attrs:
self.attribute = self.attrs[0]
if domain.class_var and domain.class_var.is_discrete:
self.group_var = domain.class_var
else:
self.group_var = None # Reset to trigger selection via callback
def apply_sorting(self):
def compute_score(attr):
if attr is group_var:
return 3
if attr.is_continuous:
# One-way ANOVA
col = data.get_column_view(attr)[0].astype(float)
groups = (col[group_col == i] for i in range(n_groups))
groups = (col[~np.isnan(col)] for col in groups)
groups = [group for group in groups if len(group)]
p = f_oneway(*groups)[1] if len(groups) > 1 else 2
else:
# Chi-square with the given distribution into groups
# (see degrees of freedom in computation of the p-value)
if not attr.values or not group_var.values:
return 2
observed = np.array(
contingency.get_contingency(data, group_var, attr))
observed = observed[observed.sum(axis=1) != 0, :]
observed = observed[:, observed.sum(axis=0) != 0]
if min(observed.shape) < 2:
return 2
expected = \
np.outer(observed.sum(axis=1), observed.sum(axis=0)) / \
np.sum(observed)
p = chisquare(observed.ravel(),
f_exp=expected.ravel(),
ddof=n_groups - 1)[1]
if math.isnan(p):
return 2
return p
data = self.dataset
if data is None:
return
domain = data.domain
attribute = self.attribute
group_var = self.group_var
if self.order_by_importance and group_var is not None:
n_groups = len(group_var.values)
group_col = data.get_column_view(group_var)[0] if \
domain.has_continuous_attributes(
include_class=True, include_metas=True) else None
self.attrs.sort(key=compute_score)
else:
self.reset_attrs(domain)
self.attribute = attribute
def reset_all_data(self):
self.clear_scene()
self.infot1.setText("")
self.attrs.clear()
self.group_vars.set_domain(None)
self.group_view.setEnabled(False)
self.is_continuous = False
self.update_display_box()
def grouping_changed(self):
self.cb_order.setEnabled(self.group_var is not None)
self.apply_sorting()
self.attr_changed()
def select_box_items(self):
temp_cond = self.conditions.copy()
for box in self.box_scene.items():
if isinstance(box, FilterGraphicsRectItem):
box.setSelected(
box.filter.conditions in [c.conditions for c in temp_cond])
def attr_changed(self):
self.compute_box_data()
self.update_display_box()
self.layout_changed()
if self.is_continuous:
heights = 90 if self.show_annotations else 60
self.box_view.centerOn(self.scene_min_x + self.scene_width / 2,
-30 - len(self.stats) * heights / 2 + 45)
else:
self.box_view.centerOn(self.scene_width / 2,
-30 - len(self.boxes) * 40 / 2 + 45)
def compute_box_data(self):
attr = self.attribute
if not attr:
return
dataset = self.dataset
self.is_continuous = attr.is_continuous
if dataset is None or not self.is_continuous and not attr.values or \
self.group_var and not self.group_var.values:
self.stats = self.dist = self.conts = []
return
if self.group_var:
self.dist = []
self.conts = contingency.get_contingency(dataset, attr,
self.group_var)
if self.is_continuous:
stats, label_texts = [], []
for i, cont in enumerate(self.conts):
if np.sum(cont[1]):
stats.append(BoxData(cont, attr, i, self.group_var))
label_texts.append(self.group_var.values[i])
self.stats = stats
self.label_txts_all = label_texts
else:
self.label_txts_all = \
[v for v, c in zip(self.group_var.values, self.conts)
if np.sum(c) > 0]
else:
self.dist = distribution.get_distribution(dataset, attr)
self.conts = []
if self.is_continuous:
self.stats = [BoxData(self.dist, attr, None)]
self.label_txts_all = [""]
self.label_txts = [
txts for stat, txts in zip(self.stats, self.label_txts_all)
if stat.n > 0
]
self.stats = [stat for stat in self.stats if stat.n > 0]
def update_display_box(self):
if self.is_continuous:
self.stretching_box.hide()
self.display_box.show()
self.compare_rb.setEnabled(self.group_var is not None)
else:
self.stretching_box.show()
self.display_box.hide()
self.sort_cb.setEnabled(self.group_var is not None)
def clear_scene(self):
self.closeContext()
self.box_scene.clearSelection()
self.box_scene.clear()
self.box_view.viewport().update()
self.attr_labels = []
self.labels = []
self.boxes = []
self.mean_labels = []
self.posthoc_lines = []
self.openContext(self.dataset)
def layout_changed(self):
attr = self.attribute
if not attr:
return
self.clear_scene()
if self.dataset is None or len(self.conts) == len(self.dist) == 0:
return
if not self.is_continuous:
self.display_changed_disc()
return
self.mean_labels = [
self.mean_label(stat, attr, lab)
for stat, lab in zip(self.stats, self.label_txts)
]
self.draw_axis()
self.boxes = [self.box_group(stat) for stat in self.stats]
self.labels = [
self.label_group(stat, attr, mean_lab)
for stat, mean_lab in zip(self.stats, self.mean_labels)
]
self.attr_labels = [
QGraphicsSimpleTextItem(lab) for lab in self.label_txts
]
for it in chain(self.labels, self.attr_labels):
self.box_scene.addItem(it)
self.display_changed()
def display_changed(self):
if self.dataset is None:
return
if not self.is_continuous:
self.display_changed_disc()
return
self.order = list(range(len(self.stats)))
criterion = self._sorting_criteria_attrs[self.compare]
if criterion:
vals = [getattr(stat, criterion) for stat in self.stats]
overmax = max((val for val in vals if val is not None), default=0) \
+ 1
vals = [val if val is not None else overmax for val in vals]
self.order = sorted(self.order, key=vals.__getitem__)
heights = 90 if self.show_annotations else 60
for row, box_index in enumerate(self.order):
y = (-len(self.stats) + row) * heights + 10
for item in self.boxes[box_index]:
self.box_scene.addItem(item)
item.setY(y)
labels = self.labels[box_index]
if self.show_annotations:
labels.show()
labels.setY(y)
else:
labels.hide()
label = self.attr_labels[box_index]
label.setY(y - 15 - label.boundingRect().height())
if self.show_annotations:
label.hide()
else:
stat = self.stats[box_index]
if self.compare == OWBoxPlot.CompareMedians and \
stat.median is not None:
pos = stat.median + 5 / self.scale_x
elif self.compare == OWBoxPlot.CompareMeans or stat.q25 is None:
pos = stat.mean + 5 / self.scale_x
else:
pos = stat.q25
label.setX(pos * self.scale_x)
label.show()
r = QRectF(self.scene_min_x, -30 - len(self.stats) * heights,
self.scene_width,
len(self.stats) * heights + 90)
self.box_scene.setSceneRect(r)
self.compute_tests()
self.show_posthoc()
self.select_box_items()
def display_changed_disc(self):
assert not self.is_continuous
self.clear_scene()
self.attr_labels = [
QGraphicsSimpleTextItem(lab) for lab in self.label_txts_all
]
if not self.stretched:
if self.group_var:
self.labels = [
QGraphicsTextItem("{}".format(int(sum(cont))))
for cont in self.conts if np.sum(cont) > 0
]
else:
self.labels = [QGraphicsTextItem(str(int(sum(self.dist))))]
self.order = list(range(len(self.attr_labels)))
self.draw_axis_disc()
if self.group_var:
self.boxes = \
[self.strudel(cont, i) for i, cont in enumerate(self.conts)
if np.sum(cont) > 0]
self.conts = self.conts[np.sum(np.array(self.conts), axis=1) > 0]
if self.sort_freqs:
# pylint: disable=invalid-unary-operand-type
self.order = sorted(
self.order, key=(-np.sum(self.conts, axis=1)).__getitem__)
else:
self.boxes = [self.strudel(self.dist)]
for row, box_index in enumerate(self.order):
y = (-len(self.boxes) + row) * 40 + 10
box = self.boxes[box_index]
bars, labels = box[::2], box[1::2]
self.__draw_group_labels(y, box_index)
if not self.stretched:
self.__draw_row_counts(y, box_index)
if self.show_labels and self.attribute is not self.group_var:
self.__draw_bar_labels(y, bars, labels)
self.__draw_bars(y, bars)
self.box_scene.setSceneRect(-self.label_width - 5,
-30 - len(self.boxes) * 40,
self.scene_width,
len(self.boxes * 40) + 90)
self.infot1.setText("")
self.select_box_items()
def __draw_group_labels(self, y, row):
"""Draw group labels
Parameters
----------
y: int
vertical offset of bars
row: int
row index
"""
label = self.attr_labels[row]
b = label.boundingRect()
label.setPos(-b.width() - 10, y - b.height() / 2)
self.box_scene.addItem(label)
def __draw_row_counts(self, y, row):
"""Draw row counts
Parameters
----------
y: int
vertical offset of bars
row: int
row index
"""
assert not self.is_continuous
label = self.labels[row]
b = label.boundingRect()
if self.group_var:
right = self.scale_x * sum(self.conts[row])
else:
right = self.scale_x * sum(self.dist)
label.setPos(right + 10, y - b.height() / 2)
self.box_scene.addItem(label)
def __draw_bar_labels(self, y, bars, labels):
"""Draw bar labels
Parameters
----------
y: int
vertical offset of bars
bars: List[FilterGraphicsRectItem]
list of bars being drawn
labels: List[QGraphicsTextItem]
list of labels for corresponding bars
"""
label = bar_part = None
for text_item, bar_part in zip(labels, bars):
label = self.Label(text_item.toPlainText())
label.setPos(bar_part.boundingRect().x(),
y - label.boundingRect().height() - 8)
label.setMaxWidth(bar_part.boundingRect().width())
self.box_scene.addItem(label)
def __draw_bars(self, y, bars):
"""Draw bars
Parameters
----------
y: int
vertical offset of bars
bars: List[FilterGraphicsRectItem]
list of bars to draw
"""
for item in bars:
item.setPos(0, y)
self.box_scene.addItem(item)
# noinspection PyPep8Naming
def compute_tests(self):
# The t-test and ANOVA are implemented here since they efficiently use
# the widget-specific data in self.stats.
# The non-parametric tests can't do this, so we use statistics.tests
def stat_ttest():
d1, d2 = self.stats
if d1.n == 0 or d2.n == 0:
return np.nan, np.nan
pooled_var = d1.var / d1.n + d2.var / d2.n
df = pooled_var ** 2 / \
((d1.var / d1.n) ** 2 / (d1.n - 1) +
(d2.var / d2.n) ** 2 / (d2.n - 1))
if pooled_var == 0:
return np.nan, np.nan
t = abs(d1.mean - d2.mean) / math.sqrt(pooled_var)
p = 2 * (1 - scipy.special.stdtr(df, t))
return t, p
# TODO: Check this function
# noinspection PyPep8Naming
def stat_ANOVA():
if any(stat.n == 0 for stat in self.stats):
return np.nan, np.nan
n = sum(stat.n for stat in self.stats)
grand_avg = sum(stat.n * stat.mean for stat in self.stats) / n
var_between = sum(stat.n * (stat.mean - grand_avg)**2
for stat in self.stats)
df_between = len(self.stats) - 1
var_within = sum(stat.n * stat.var for stat in self.stats)
df_within = n - len(self.stats)
F = (var_between / df_between) / (var_within / df_within)
p = 1 - scipy.special.fdtr(df_between, df_within, F)
return F, p
if self.compare == OWBoxPlot.CompareNone or len(self.stats) < 2:
t = ""
elif any(s.n <= 1 for s in self.stats):
t = "At least one group has just one instance," \
"cannot compute significance"
elif len(self.stats) == 2:
if self.compare == OWBoxPlot.CompareMedians:
t = ""
# z, self.p = tests.wilcoxon_rank_sum(
# self.stats[0].dist, self.stats[1].dist)
# t = "Mann-Whitney's z: %.1f (p=%.3f)" % (z, self.p)
else:
t, self.p = stat_ttest()
t = "Student's t: %.3f (p=%.3f)" % (t, self.p)
else:
if self.compare == OWBoxPlot.CompareMedians:
t = ""
# U, self.p = -1, -1
# t = "Kruskal Wallis's U: %.1f (p=%.3f)" % (U, self.p)
else:
F, self.p = stat_ANOVA()
t = "ANOVA: %.3f (p=%.3f)" % (F, self.p)
self.infot1.setText("<center>%s</center>" % t)
def mean_label(self, stat, attr, val_name):
label = QGraphicsItemGroup()
t = QGraphicsSimpleTextItem(
"%.*f" % (attr.number_of_decimals + 1, stat.mean), label)
t.setFont(self._label_font)
bbox = t.boundingRect()
w2, h = bbox.width() / 2, bbox.height()
t.setPos(-w2, -h)
tpm = QGraphicsSimpleTextItem(
" \u00b1 " + "%.*f" % (attr.number_of_decimals + 1, stat.dev),
label)
tpm.setFont(self._label_font)
tpm.setPos(w2, -h)
if val_name:
vnm = QGraphicsSimpleTextItem(val_name + ": ", label)
vnm.setFont(self._label_font)
vnm.setBrush(self._attr_brush)
vb = vnm.boundingRect()
label.min_x = -w2 - vb.width()
vnm.setPos(label.min_x, -h)
else:
label.min_x = -w2
return label
def draw_axis(self):
"""Draw the horizontal axis and sets self.scale_x"""
misssing_stats = not self.stats
stats = self.stats or [BoxData(np.array([[0.], [1.]]), self.attribute)]
mean_labels = self.mean_labels or [
self.mean_label(stats[0], self.attribute, "")
]
bottom = min(stat.a_min for stat in stats)
top = max(stat.a_max for stat in stats)
first_val, step = compute_scale(bottom, top)
while bottom <= first_val:
first_val -= step
bottom = first_val
no_ticks = math.ceil((top - first_val) / step) + 1
top = max(top, first_val + no_ticks * step)
gbottom = min(bottom, min(stat.mean - stat.dev for stat in stats))
gtop = max(top, max(stat.mean + stat.dev for stat in stats))
bv = self.box_view
viewrect = bv.viewport().rect().adjusted(15, 15, -15, -30)
self.scale_x = scale_x = viewrect.width() / (gtop - gbottom)
# In principle we should repeat this until convergence since the new
# scaling is too conservative. (No chance am I doing this.)
mlb = min(stat.mean + mean_lab.min_x / scale_x
for stat, mean_lab in zip(stats, mean_labels))
if mlb < gbottom:
gbottom = mlb
self.scale_x = scale_x = viewrect.width() / (gtop - gbottom)
self.scene_min_x = gbottom * scale_x
self.scene_width = (gtop - gbottom) * scale_x
val = first_val
decimals = max(3, 4 - int(math.log10(step)))
while True:
l = self.box_scene.addLine(val * scale_x, -1, val * scale_x, 1,
self._pen_axis_tick)
l.setZValue(100)
t = self.box_scene.addSimpleText(
repr(round(val, decimals)) if not misssing_stats else "?",
self._axis_font)
t.setFlags(t.flags() | QGraphicsItem.ItemIgnoresTransformations)
r = t.boundingRect()
t.setPos(val * scale_x - r.width() / 2, 8)
if val >= top:
break
val += step
self.box_scene.addLine(bottom * scale_x - 4, 0, top * scale_x + 4, 0,
self._pen_axis)
def draw_axis_disc(self):
"""
Draw the horizontal axis and sets self.scale_x for discrete attributes
"""
assert not self.is_continuous
if self.stretched:
if not self.attr_labels:
return
step = steps = 10
else:
if self.group_var:
max_box = max(float(np.sum(dist)) for dist in self.conts)
else:
max_box = float(np.sum(self.dist))
if max_box == 0:
self.scale_x = 1
return
_, step = compute_scale(0, max_box)
step = int(step) if step > 1 else 1
steps = int(math.ceil(max_box / step))
max_box = step * steps
bv = self.box_view
viewrect = bv.viewport().rect().adjusted(15, 15, -15, -30)
self.scene_width = viewrect.width()
lab_width = max(lab.boundingRect().width() for lab in self.attr_labels)
lab_width = max(lab_width, 40)
lab_width = min(lab_width, self.scene_width / 3)
self.label_width = lab_width
right_offset = 0 # offset for the right label
if not self.stretched and self.labels:
if self.group_var:
rows = list(zip(self.conts, self.labels))
else:
rows = [(self.dist, self.labels[0])]
# available space left of the 'group labels'
available = self.scene_width - lab_width - 10
scale_x = (available - right_offset) / max_box
max_right = max(
sum(dist) * scale_x + 10 + lbl.boundingRect().width()
for dist, lbl in rows)
right_offset = max(0, max_right - max_box * scale_x)
self.scale_x = scale_x = \
(self.scene_width - lab_width - 10 - right_offset) / max_box
self.box_scene.addLine(0, 0, max_box * scale_x, 0, self._pen_axis)
for val in range(0, step * steps + 1, step):
l = self.box_scene.addLine(val * scale_x, -1, val * scale_x, 1,
self._pen_axis_tick)
l.setZValue(100)
t = self.box_scene.addSimpleText(str(val), self._axis_font)
t.setPos(val * scale_x - t.boundingRect().width() / 2, 8)
if self.stretched:
self.scale_x *= 100
def label_group(self, stat, attr, mean_lab):
def centered_text(val, pos):
t = QGraphicsSimpleTextItem(
"%.*f" % (attr.number_of_decimals + 1, val), labels)
t.setFont(self._label_font)
bbox = t.boundingRect()
t.setPos(pos - bbox.width() / 2, 22)
return t
def line(x, down=1):
QGraphicsLineItem(x, 12 * down, x, 20 * down, labels)
def move_label(label, frm, to):
label.setX(to)
to += t_box.width() / 2
path = QPainterPath()
path.lineTo(0, 4)
path.lineTo(to - frm, 4)
path.lineTo(to - frm, 8)
p = QGraphicsPathItem(path)
p.setPos(frm, 12)
labels.addToGroup(p)
labels = QGraphicsItemGroup()
labels.addToGroup(mean_lab)
m = stat.mean * self.scale_x
mean_lab.setPos(m, -22)
line(m, -1)
if stat.median is not None:
msc = stat.median * self.scale_x
med_t = centered_text(stat.median, msc)
med_box_width2 = med_t.boundingRect().width() / 2
line(msc)
if stat.q25 is not None:
x = stat.q25 * self.scale_x
t = centered_text(stat.q25, x)
t_box = t.boundingRect()
med_left = msc - med_box_width2
if x + t_box.width() / 2 >= med_left - 5:
move_label(t, x, med_left - t_box.width() - 5)
else:
line(x)
if stat.q75 is not None:
x = stat.q75 * self.scale_x
t = centered_text(stat.q75, x)
t_box = t.boundingRect()
med_right = msc + med_box_width2
if x - t_box.width() / 2 <= med_right + 5:
move_label(t, x, med_right + 5)
else:
line(x)
return labels
def box_group(self, stat, height=20):
def line(x0, y0, x1, y1, *args):
return QGraphicsLineItem(x0 * scale_x, y0, x1 * scale_x, y1, *args)
scale_x = self.scale_x
box = []
whisker1 = line(stat.a_min, -1.5, stat.a_min, 1.5)
whisker2 = line(stat.a_max, -1.5, stat.a_max, 1.5)
vert_line = line(stat.a_min, 0, stat.a_max, 0)
mean_line = line(stat.mean, -height / 3, stat.mean, height / 3)
for it in (whisker1, whisker2, mean_line):
it.setPen(self._pen_paramet)
vert_line.setPen(self._pen_dotted)
var_line = line(stat.mean - stat.dev, 0, stat.mean + stat.dev, 0)
var_line.setPen(self._pen_paramet)
box.extend([whisker1, whisker2, vert_line, mean_line, var_line])
if stat.q25 is not None and stat.q75 is not None:
mbox = FilterGraphicsRectItem(stat.conditions, stat.q25 * scale_x,
-height / 2,
(stat.q75 - stat.q25) * scale_x,
height)
mbox.setBrush(self._box_brush)
mbox.setPen(QPen(Qt.NoPen))
mbox.setZValue(-200)
box.append(mbox)
if stat.median is not None:
median_line = line(stat.median, -height / 2, stat.median,
height / 2)
median_line.setPen(self._pen_median)
median_line.setZValue(-150)
box.append(median_line)
return box
def strudel(self, dist, group_val_index=None):
attr = self.attribute
ss = np.sum(dist)
box = []
if ss < 1e-6:
cond = [FilterDiscrete(attr, None)]
if group_val_index is not None:
cond.append(FilterDiscrete(self.group_var, [group_val_index]))
box.append(FilterGraphicsRectItem(cond, 0, -10, 1, 10))
cum = 0
for i, v in enumerate(dist):
if v < 1e-6:
continue
if self.stretched:
v /= ss
v *= self.scale_x
cond = [FilterDiscrete(attr, [i])]
if group_val_index is not None:
cond.append(FilterDiscrete(self.group_var, [group_val_index]))
rect = FilterGraphicsRectItem(cond, cum + 1, -6, v - 2, 12)
rect.setBrush(QBrush(QColor(*attr.colors[i])))
rect.setPen(QPen(Qt.NoPen))
if self.stretched:
tooltip = "{}: {:.2f}%".format(attr.values[i],
100 * dist[i] / sum(dist))
else:
tooltip = "{}: {}".format(attr.values[i], int(dist[i]))
rect.setToolTip(tooltip)
text = QGraphicsTextItem(attr.values[i])
box.append(rect)
box.append(text)
cum += v
return box
def commit(self):
self.conditions = [
item.filter for item in self.box_scene.selectedItems()
if item.filter
]
selected, selection = None, []
if self.conditions:
selected = Values(self.conditions, conjunction=False)(self.dataset)
selection = np.in1d(self.dataset.ids,
selected.ids,
assume_unique=True).nonzero()[0]
self.Outputs.selected_data.send(selected)
self.Outputs.annotated_data.send(
create_annotated_table(self.dataset, selection))
def show_posthoc(self):
def line(y0, y1):
it = self.box_scene.addLine(x, y0, x, y1, self._post_line_pen)
it.setZValue(-100)
self.posthoc_lines.append(it)
while self.posthoc_lines:
self.box_scene.removeItem(self.posthoc_lines.pop())
if self.compare == OWBoxPlot.CompareNone or len(self.stats) < 2:
return
if self.compare == OWBoxPlot.CompareMedians:
crit_line = "median"
else:
crit_line = "mean"
xs = []
height = 90 if self.show_annotations else 60
y_up = -len(self.stats) * height + 10
for pos, box_index in enumerate(self.order):
stat = self.stats[box_index]
x = getattr(stat, crit_line)
if x is None:
continue
x *= self.scale_x
xs.append(x * self.scale_x)
by = y_up + pos * height
line(by + 12, 3)
line(by - 12, by - 25)
used_to = []
last_to = to = 0
for frm, frm_x in enumerate(xs[:-1]):
for to in range(frm + 1, len(xs)):
if xs[to] - frm_x > 1.5:
to -= 1
break
if to in (last_to, frm):
continue
for rowi, used in enumerate(used_to):
if used < frm:
used_to[rowi] = to
break
else:
rowi = len(used_to)
used_to.append(to)
y = -6 - rowi * 6
it = self.box_scene.addLine(frm_x - 2, y, xs[to] + 2, y,
self._post_grp_pen)
self.posthoc_lines.append(it)
last_to = to
def get_widget_name_extension(self):
return self.attribute.name if self.attribute else None
def send_report(self):
self.report_plot()
text = ""
if self.attribute:
text += "Box plot for attribute '{}' ".format(self.attribute.name)
if self.group_var:
text += "grouped by '{}'".format(self.group_var.name)
if text:
self.report_caption(text)
class Label(QGraphicsSimpleTextItem):
"""Boxplot Label with settable maxWidth"""
# Minimum width to display label text
MIN_LABEL_WIDTH = 25
# padding bellow the text
PADDING = 3
__max_width = None
def maxWidth(self):
return self.__max_width
def setMaxWidth(self, max_width):
self.__max_width = max_width
def paint(self, painter, option, widget):
"""Overrides QGraphicsSimpleTextItem.paint
If label text is too long, it is elided
to fit into the allowed region
"""
if self.__max_width is None:
width = option.rect.width()
else:
width = self.__max_width
if width < self.MIN_LABEL_WIDTH:
# if space is too narrow, no label
return
fm = painter.fontMetrics()
text = fm.elidedText(self.text(), Qt.ElideRight, width)
painter.drawText(
option.rect.x(),
option.rect.y() + self.boundingRect().height() - self.PADDING,
text)
return QSizePolicy(QSizePolicy.Expanding, QSizePolicy.Expanding)
if __name__ == '__main__':
import sys
from Orange.data.table import Table
from AnyQt.QtWidgets import ( # pylint: disable=ungrouped-imports
QGraphicsView, QGraphicsScene, QApplication, QWidget)
app = QApplication(sys.argv)
widget = QWidget()
widget.resize(500, 300)
scene = QGraphicsScene(widget)
view = QGraphicsView(scene, widget)
dataset = Table(sys.argv[1] if len(sys.argv) > 1 else 'iris')
histogram = Histogram(
dataset,
variable=0,
height=300,
width=500,
n_bins=20,
bar_spacing=2,
border=(0, 0, 5, 0),
border_color='#000',
color_attribute='iris',
)
scene.addItem(histogram)
widget.show()
app.exec()
def data(self, index, role):
# type: (QModelIndex, Qt.ItemDataRole) -> Any
# Text formatting for various data simply requires a lot of branches.
# This is much better than overengineering various formatters...
# pylint: disable=too-many-branches
if not index.isValid():
return None
row, column = self.mapToSourceRows(index.row()), index.column()
# Make sure we're not out of range
if not 0 <= row <= self.n_attributes:
return QVariant()
attribute = self.variables[row]
if role == Qt.BackgroundRole:
if attribute in self.domain.attributes:
return self.COLOR_FOR_ROLE[self.ATTRIBUTE]
elif attribute in self.domain.metas:
return self.COLOR_FOR_ROLE[self.META]
elif attribute in self.domain.class_vars:
return self.COLOR_FOR_ROLE[self.CLASS_VAR]
elif role == Qt.TextAlignmentRole:
if column == self.Columns.NAME:
return Qt.AlignLeft | Qt.AlignVCenter
return Qt.AlignRight | Qt.AlignVCenter
output = None
if column == self.Columns.ICON:
if role == Qt.DecorationRole:
return gui.attributeIconDict[attribute]
elif column == self.Columns.NAME:
if role == Qt.DisplayRole:
output = attribute.name
elif column == self.Columns.DISTRIBUTION:
if role == Qt.DisplayRole:
if isinstance(attribute,
(DiscreteVariable, ContinuousVariable)):
if row not in self.__distributions_cache:
scene = QGraphicsScene(parent=self)
histogram = Histogram(
data=self.table,
variable=attribute,
color_attribute=self.target_var,
border=(0, 0, 2, 0),
border_color='#ccc',
)
scene.addItem(histogram)
self.__distributions_cache[row] = scene
return self.__distributions_cache[row]
elif column == self.Columns.CENTER:
if role == Qt.DisplayRole:
if isinstance(attribute, DiscreteVariable):
output = self._center[row]
if not np.isnan(output):
output = attribute.str_val(self._center[row])
elif isinstance(attribute, TimeVariable):
output = attribute.str_val(self._center[row])
else:
output = self._center[row]
elif column == self.Columns.DISPERSION:
if role == Qt.DisplayRole:
if isinstance(attribute, TimeVariable):
output = format_time_diff(self._min[row], self._max[row])
else:
output = self._dispersion[row]
elif column == self.Columns.MIN:
if role == Qt.DisplayRole:
if isinstance(attribute, DiscreteVariable):
if attribute.ordered:
output = attribute.str_val(self._min[row])
elif isinstance(attribute, TimeVariable):
output = attribute.str_val(self._min[row])
else:
output = self._min[row]
elif column == self.Columns.MAX:
if role == Qt.DisplayRole:
if isinstance(attribute, DiscreteVariable):
if attribute.ordered:
output = attribute.str_val(self._max[row])
elif isinstance(attribute, TimeVariable):
output = attribute.str_val(self._max[row])
else:
output = self._max[row]
elif column == self.Columns.MISSING:
if role == Qt.DisplayRole:
output = '%d (%d%%)' % (self._missing[row], 100 *
self._missing[row] / self.n_instances)
# Consistently format the text inside the table cells
# The easiest way to check for NaN is to compare with itself
if output != output: # pylint: disable=comparison-with-itself
output = ''
# Format ∞ properly
elif output in (np.inf, -np.inf):
output = '%s∞' % ['', '-'][output < 0]
elif isinstance(output, int):
output = locale.format_string('%d', output, grouping=True)
elif isinstance(output, float):
output = locale.format_string('%.2f', output, grouping=True)
return output
class OWNomogram(OWWidget):
name = "Nomogram"
description = " Nomograms for Visualization of Naive Bayesian" \
" and Logistic Regression Classifiers."
icon = "icons/Nomogram.svg"
priority = 2000
inputs = [("Classifier", Model, "set_classifier"),
("Data", Table, "set_instance")]
MAX_N_ATTRS = 1000
POINT_SCALE = 0
ALIGN_LEFT = 0
ALIGN_ZERO = 1
ACCEPTABLE = (NaiveBayesModel, LogisticRegressionClassifier)
settingsHandler = ClassValuesContextHandler()
target_class_index = ContextSetting(0)
normalize_probabilities = Setting(False)
scale = Setting(1)
display_index = Setting(1)
n_attributes = Setting(10)
sort_index = Setting(SortBy.ABSOLUTE)
cont_feature_dim_index = Setting(0)
graph_name = "scene"
class Error(OWWidget.Error):
invalid_classifier = Msg("Nomogram accepts only Naive Bayes and "
"Logistic Regression classifiers.")
def __init__(self):
super().__init__()
self.instances = None
self.domain = None
self.data = None
self.classifier = None
self.align = OWNomogram.ALIGN_ZERO
self.log_odds_ratios = []
self.log_reg_coeffs = []
self.log_reg_coeffs_orig = []
self.log_reg_cont_data_extremes = []
self.p = None
self.b0 = None
self.points = []
self.feature_items = []
self.feature_marker_values = []
self.scale_back = lambda x: x
self.scale_forth = lambda x: x
self.nomogram = None
self.nomogram_main = None
self.vertical_line = None
self.hidden_vertical_line = None
self.old_target_class_index = self.target_class_index
self.markers_set = False
self.repaint = False
# GUI
box = gui.vBox(self.controlArea, "Target class")
self.class_combo = gui.comboBox(box,
self,
"target_class_index",
callback=self._class_combo_changed,
contentsLength=12)
self.norm_check = gui.checkBox(
box,
self,
"normalize_probabilities",
"Normalize probabilities",
hidden=True,
callback=self._norm_check_changed,
tooltip="For multiclass data 1 vs. all probabilities do not"
" sum to 1 and therefore could be normalized.")
self.scale_radio = gui.radioButtons(
self.controlArea,
self,
"scale", ["Point scale", "Log odds ratios"],
box="Scale",
callback=self._radio_button_changed)
box = gui.vBox(self.controlArea, "Display features")
grid = QGridLayout()
self.display_radio = gui.radioButtonsInBox(
box,
self,
"display_index", [],
orientation=grid,
callback=self._display_radio_button_changed)
radio_all = gui.appendRadioButton(self.display_radio,
"All:",
addToLayout=False)
radio_best = gui.appendRadioButton(self.display_radio,
"Best ranked:",
addToLayout=False)
spin_box = gui.hBox(None, margin=0)
self.n_spin = gui.spin(spin_box,
self,
"n_attributes",
1,
self.MAX_N_ATTRS,
label=" ",
controlWidth=60,
callback=self._n_spin_changed)
grid.addWidget(radio_all, 1, 1)
grid.addWidget(radio_best, 2, 1)
grid.addWidget(spin_box, 2, 2)
self.sort_combo = gui.comboBox(box,
self,
"sort_index",
label="Sort by: ",
items=SortBy.items(),
orientation=Qt.Horizontal,
callback=self._sort_combo_changed)
self.cont_feature_dim_combo = gui.comboBox(
box,
self,
"cont_feature_dim_index",
label="Continuous features: ",
items=["1D projection", "2D curve"],
orientation=Qt.Horizontal,
callback=self._cont_feature_dim_combo_changed)
gui.rubber(self.controlArea)
self.scene = QGraphicsScene()
self.view = QGraphicsView(
self.scene,
horizontalScrollBarPolicy=Qt.ScrollBarAlwaysOff,
renderHints=QPainter.Antialiasing | QPainter.TextAntialiasing
| QPainter.SmoothPixmapTransform,
alignment=Qt.AlignLeft)
self.view.viewport().installEventFilter(self)
self.view.viewport().setMinimumWidth(300)
self.view.sizeHint = lambda: QSize(600, 500)
self.mainArea.layout().addWidget(self.view)
def _class_combo_changed(self):
values = [item.dot.value for item in self.feature_items]
self.feature_marker_values = self.scale_back(values)
coeffs = [
np.nan_to_num(p[self.target_class_index] /
p[self.old_target_class_index]) for p in self.points
]
points = [p[self.old_target_class_index] for p in self.points]
self.feature_marker_values = [
self.get_points_from_coeffs(v, c, p)
for (v, c, p) in zip(self.feature_marker_values, coeffs, points)
]
self.update_scene()
self.old_target_class_index = self.target_class_index
def _norm_check_changed(self):
values = [item.dot.value for item in self.feature_items]
self.feature_marker_values = self.scale_back(values)
self.update_scene()
def _radio_button_changed(self):
values = [item.dot.value for item in self.feature_items]
self.feature_marker_values = self.scale_back(values)
self.update_scene()
def _display_radio_button_changed(self):
self.__hide_attrs(self.n_attributes if self.display_index else None)
def _n_spin_changed(self):
self.display_index = 1
self.__hide_attrs(self.n_attributes)
def __hide_attrs(self, n_show):
if self.nomogram_main is None:
return
self.nomogram_main.hide(n_show)
if self.vertical_line:
x = self.vertical_line.line().x1()
y = self.nomogram_main.layout.preferredHeight() + 30
self.vertical_line.setLine(x, -6, x, y)
self.hidden_vertical_line.setLine(x, -6, x, y)
rect = QRectF(self.scene.sceneRect().x(),
self.scene.sceneRect().y(),
self.scene.itemsBoundingRect().width(),
self.nomogram.preferredSize().height())
self.scene.setSceneRect(rect.adjusted(0, 0, 70, 70))
def _sort_combo_changed(self):
if self.nomogram_main is None:
return
self.nomogram_main.hide(None)
self.nomogram_main.sort(self.sort_index)
self.__hide_attrs(self.n_attributes if self.display_index else None)
def _cont_feature_dim_combo_changed(self):
values = [item.dot.value for item in self.feature_items]
self.feature_marker_values = self.scale_back(values)
self.update_scene()
def eventFilter(self, obj, event):
if obj is self.view.viewport() and event.type() == QEvent.Resize:
self.repaint = True
values = [item.dot.value for item in self.feature_items]
self.feature_marker_values = self.scale_back(values)
self.update_scene()
return super().eventFilter(obj, event)
def update_controls(self):
self.class_combo.clear()
self.norm_check.setHidden(True)
self.cont_feature_dim_combo.setEnabled(True)
if self.domain:
self.class_combo.addItems(self.domain.class_vars[0].values)
if len(self.domain.attributes) > self.MAX_N_ATTRS:
self.display_index = 1
if len(self.domain.class_vars[0].values) > 2:
self.norm_check.setHidden(False)
if not self.domain.has_continuous_attributes():
self.cont_feature_dim_combo.setEnabled(False)
self.cont_feature_dim_index = 0
model = self.sort_combo.model()
item = model.item(SortBy.POSITIVE)
item.setFlags(item.flags() | Qt.ItemIsEnabled)
item = model.item(SortBy.NEGATIVE)
item.setFlags(item.flags() | Qt.ItemIsEnabled)
self.align = OWNomogram.ALIGN_ZERO
if self.classifier and isinstance(self.classifier,
LogisticRegressionClassifier):
self.align = OWNomogram.ALIGN_LEFT
item = model.item(SortBy.POSITIVE)
item.setFlags(item.flags() & ~Qt.ItemIsEnabled)
item = model.item(SortBy.NEGATIVE)
item.setFlags(item.flags() & ~Qt.ItemIsEnabled)
if self.sort_index in (SortBy.POSITIVE, SortBy.POSITIVE):
self.sort_index = SortBy.NO_SORTING
def set_instance(self, data):
self.instances = data
self.feature_marker_values = []
self.set_feature_marker_values()
def set_classifier(self, classifier):
self.closeContext()
self.classifier = classifier
self.Error.clear()
if self.classifier and not isinstance(self.classifier,
self.ACCEPTABLE):
self.Error.invalid_classifier()
self.classifier = None
self.domain = self.classifier.domain if self.classifier else None
self.data = None
self.calculate_log_odds_ratios()
self.calculate_log_reg_coefficients()
self.update_controls()
self.target_class_index = 0
self.openContext(self.domain and self.domain.class_var)
self.points = self.log_odds_ratios or self.log_reg_coeffs
self.feature_marker_values = []
self.old_target_class_index = self.target_class_index
self.update_scene()
def calculate_log_odds_ratios(self):
self.log_odds_ratios = []
self.p = None
if self.classifier is None or self.domain is None:
return
if not isinstance(self.classifier, NaiveBayesModel):
return
log_cont_prob = self.classifier.log_cont_prob
class_prob = self.classifier.class_prob
for i in range(len(self.domain.attributes)):
ca = np.exp(log_cont_prob[i]) * class_prob[:, None]
_or = (ca / (1 - ca)) / (class_prob / (1 - class_prob))[:, None]
self.log_odds_ratios.append(np.log(_or))
self.p = class_prob
def calculate_log_reg_coefficients(self):
self.log_reg_coeffs = []
self.log_reg_cont_data_extremes = []
self.b0 = None
if self.classifier is None or self.domain is None:
return
if not isinstance(self.classifier, LogisticRegressionClassifier):
return
self.domain = self.reconstruct_domain(self.classifier.original_domain,
self.domain)
self.data = self.classifier.original_data.transform(self.domain)
attrs, ranges, start = self.domain.attributes, [], 0
for attr in attrs:
stop = start + len(attr.values) if attr.is_discrete else start + 1
ranges.append(slice(start, stop))
start = stop
self.b0 = self.classifier.intercept
coeffs = self.classifier.coefficients
if len(self.domain.class_var.values) == 2:
self.b0 = np.hstack((self.b0 * (-1), self.b0))
coeffs = np.vstack((coeffs * (-1), coeffs))
self.log_reg_coeffs = [coeffs[:, ranges[i]] for i in range(len(attrs))]
self.log_reg_coeffs_orig = self.log_reg_coeffs.copy()
min_values = nanmin(self.data.X, axis=0)
max_values = nanmax(self.data.X, axis=0)
for i, min_t, max_t in zip(range(len(self.log_reg_coeffs)), min_values,
max_values):
if self.log_reg_coeffs[i].shape[1] == 1:
coef = self.log_reg_coeffs[i]
self.log_reg_coeffs[i] = np.hstack(
(coef * min_t, coef * max_t))
self.log_reg_cont_data_extremes.append(
[sorted([min_t, max_t], reverse=(c < 0)) for c in coef])
else:
self.log_reg_cont_data_extremes.append([None])
def update_scene(self):
if not self.repaint:
return
self.clear_scene()
if self.domain is None or not len(self.points[0]):
return
name_items = [
QGraphicsTextItem(a.name) for a in self.domain.attributes
]
point_text = QGraphicsTextItem("Points")
probs_text = QGraphicsTextItem("Probabilities (%)")
all_items = name_items + [point_text, probs_text]
name_offset = -max(t.boundingRect().width() for t in all_items) - 50
w = self.view.viewport().rect().width()
max_width = w + name_offset - 100
points = [pts[self.target_class_index] for pts in self.points]
minimums = [min(p) for p in points]
if self.align == OWNomogram.ALIGN_LEFT:
points = [p - m for m, p in zip(minimums, points)]
max_ = np.nan_to_num(max(max(abs(p)) for p in points))
d = 100 / max_ if max_ else 1
if self.scale == OWNomogram.POINT_SCALE:
points = [p * d for p in points]
if self.scale == OWNomogram.POINT_SCALE and \
self.align == OWNomogram.ALIGN_LEFT:
self.scale_back = lambda x: [
p / d + m for m, p in zip(minimums, x)
]
self.scale_forth = lambda x: [(p - m) * d
for m, p in zip(minimums, x)]
if self.scale == OWNomogram.POINT_SCALE and \
self.align != OWNomogram.ALIGN_LEFT:
self.scale_back = lambda x: [p / d for p in x]
self.scale_forth = lambda x: [p * d for p in x]
if self.scale != OWNomogram.POINT_SCALE and \
self.align == OWNomogram.ALIGN_LEFT:
self.scale_back = lambda x: [p + m for m, p in zip(minimums, x)]
self.scale_forth = lambda x: [p - m for m, p in zip(minimums, x)]
if self.scale != OWNomogram.POINT_SCALE and \
self.align != OWNomogram.ALIGN_LEFT:
self.scale_back = lambda x: x
self.scale_forth = lambda x: x
point_item, nomogram_head = self.create_main_nomogram(
name_items, points, max_width, point_text, name_offset)
probs_item, nomogram_foot = self.create_footer_nomogram(
probs_text, d, minimums, max_width, name_offset)
for item in self.feature_items:
item.dot.point_dot = point_item.dot
item.dot.probs_dot = probs_item.dot
item.dot.vertical_line = self.hidden_vertical_line
self.nomogram = nomogram = NomogramItem()
nomogram.add_items([nomogram_head, self.nomogram_main, nomogram_foot])
self.scene.addItem(nomogram)
self.set_feature_marker_values()
rect = QRectF(self.scene.itemsBoundingRect().x(),
self.scene.itemsBoundingRect().y(),
self.scene.itemsBoundingRect().width(),
self.nomogram.preferredSize().height())
self.scene.setSceneRect(rect.adjusted(0, 0, 70, 70))
def create_main_nomogram(self, name_items, points, max_width, point_text,
name_offset):
cls_index = self.target_class_index
min_p = min(min(p) for p in points)
max_p = max(max(p) for p in points)
values = self.get_ruler_values(min_p, max_p, max_width)
min_p, max_p = min(values), max(values)
diff_ = np.nan_to_num(max_p - min_p)
scale_x = max_width / diff_ if diff_ else max_width
nomogram_header = NomogramItem()
point_item = RulerItem(point_text, values, scale_x, name_offset,
-scale_x * min_p)
point_item.setPreferredSize(point_item.preferredWidth(), 35)
nomogram_header.add_items([point_item])
self.nomogram_main = SortableNomogramItem()
cont_feature_item_class = ContinuousFeature2DItem if \
self.cont_feature_dim_index else ContinuousFeatureItem
self.feature_items = [
DiscreteFeatureItem(name_items[i], [val for val in att.values],
points[i], scale_x, name_offset, -scale_x *
min_p, self.points[i][cls_index])
if att.is_discrete else cont_feature_item_class(
name_items[i], self.log_reg_cont_data_extremes[i][cls_index],
self.get_ruler_values(
np.min(points[i]), np.max(points[i]),
scale_x * (np.max(points[i]) - np.min(points[i])),
False), scale_x, name_offset, -scale_x *
min_p, self.log_reg_coeffs_orig[i][cls_index][0])
for i, att in enumerate(self.domain.attributes)
]
self.nomogram_main.add_items(
self.feature_items, self.sort_index,
self.n_attributes if self.display_index else None)
x = -scale_x * min_p
y = self.nomogram_main.layout.preferredHeight() + 30
self.vertical_line = QGraphicsLineItem(x, -6, x, y)
self.vertical_line.setPen(QPen(Qt.DotLine))
self.vertical_line.setParentItem(point_item)
self.hidden_vertical_line = QGraphicsLineItem(x, -6, x, y)
pen = QPen(Qt.DashLine)
pen.setBrush(QColor(Qt.red))
self.hidden_vertical_line.setPen(pen)
self.hidden_vertical_line.setParentItem(point_item)
return point_item, nomogram_header
def create_footer_nomogram(self, probs_text, d, minimums, max_width,
name_offset):
eps, d_ = 0.05, 1
k = -np.log(self.p / (1 - self.p)) if self.p is not None else -self.b0
min_sum = k[self.target_class_index] - np.log((1 - eps) / eps)
max_sum = k[self.target_class_index] - np.log(eps / (1 - eps))
if self.align == OWNomogram.ALIGN_LEFT:
max_sum = max_sum - sum(minimums)
min_sum = min_sum - sum(minimums)
for i in range(len(k)):
k[i] = k[i] - sum(
[min(q) for q in [p[i] for p in self.points]])
if self.scale == OWNomogram.POINT_SCALE:
min_sum *= d
max_sum *= d
d_ = d
values = self.get_ruler_values(min_sum, max_sum, max_width)
min_sum, max_sum = min(values), max(values)
diff_ = np.nan_to_num(max_sum - min_sum)
scale_x = max_width / diff_ if diff_ else max_width
cls_var, cls_index = self.domain.class_var, self.target_class_index
nomogram_footer = NomogramItem()
def get_normalized_probabilities(val):
if not self.normalize_probabilities:
return 1 / (1 + np.exp(k[cls_index] - val / d_))
totals = self.__get_totals_for_class_values(minimums)
p_sum = np.sum(1 / (1 + np.exp(k - totals / d_)))
return 1 / (1 + np.exp(k[cls_index] - val / d_)) / p_sum
def get_points(prob):
if not self.normalize_probabilities:
return (k[cls_index] - np.log(1 / prob - 1)) * d_
totals = self.__get_totals_for_class_values(minimums)
p_sum = np.sum(1 / (1 + np.exp(k - totals / d_)))
return (k[cls_index] - np.log(1 / (prob * p_sum) - 1)) * d_
self.markers_set = False
probs_item = ProbabilitiesRulerItem(
probs_text,
values,
scale_x,
name_offset,
-scale_x * min_sum,
get_points=get_points,
title="{}='{}'".format(cls_var.name, cls_var.values[cls_index]),
get_probabilities=get_normalized_probabilities)
self.markers_set = True
nomogram_footer.add_items([probs_item])
return probs_item, nomogram_footer
def __get_totals_for_class_values(self, minimums):
cls_index = self.target_class_index
marker_values = [item.dot.value for item in self.feature_items]
if not self.markers_set:
marker_values = self.scale_forth(marker_values)
totals = np.empty(len(self.domain.class_var.values))
totals[cls_index] = sum(marker_values)
marker_values = self.scale_back(marker_values)
for i in range(len(self.domain.class_var.values)):
if i == cls_index:
continue
coeffs = [np.nan_to_num(p[i] / p[cls_index]) for p in self.points]
points = [p[cls_index] for p in self.points]
total = sum([
self.get_points_from_coeffs(v, c, p)
for (v, c, p) in zip(marker_values, coeffs, points)
])
if self.align == OWNomogram.ALIGN_LEFT:
points = [p - m for m, p in zip(minimums, points)]
total -= sum([min(p) for p in [p[i] for p in self.points]])
d = 100 / max(max(abs(p)) for p in points)
if self.scale == OWNomogram.POINT_SCALE:
total *= d
totals[i] = total
return totals
def set_feature_marker_values(self):
if not (len(self.points) and len(self.feature_items)):
return
if not len(self.feature_marker_values):
self._init_feature_marker_values()
self.feature_marker_values = self.scale_forth(
self.feature_marker_values)
item = self.feature_items[0]
for i, item in enumerate(self.feature_items):
item.dot.move_to_val(self.feature_marker_values[i])
item.dot.probs_dot.move_to_sum()
def _init_feature_marker_values(self):
self.feature_marker_values = []
cls_index = self.target_class_index
instances = Table(self.domain, self.instances) \
if self.instances else None
for i, attr in enumerate(self.domain.attributes):
value, feature_val = 0, None
if len(self.log_reg_coeffs):
if attr.is_discrete:
ind, n = unique(self.data.X[:, i], return_counts=True)
feature_val = np.nan_to_num(ind[np.argmax(n)])
else:
feature_val = mean(self.data.X[:, i])
inst_in_dom = instances and attr in instances.domain
if inst_in_dom and not np.isnan(instances[0][attr]):
feature_val = instances[0][attr]
if feature_val is not None:
value = self.points[i][cls_index][int(feature_val)] \
if attr.is_discrete else \
self.log_reg_coeffs_orig[i][cls_index][0] * feature_val
self.feature_marker_values.append(value)
def clear_scene(self):
self.feature_items = []
self.scale_back = lambda x: x
self.scale_forth = lambda x: x
self.nomogram = None
self.nomogram_main = None
self.vertical_line = None
self.hidden_vertical_line = None
self.scene.clear()
def send_report(self):
self.report_plot()
@staticmethod
def reconstruct_domain(original, preprocessed):
# abuse dict to make "in" comparisons faster
attrs = OrderedDict()
for attr in preprocessed.attributes:
cv = attr._compute_value.variable._compute_value
var = cv.variable if cv else original[attr.name]
if var in attrs: # the reason for OrderedDict
continue
attrs[var] = None # we only need keys
attrs = list(attrs.keys())
return Domain(attrs, original.class_var, original.metas)
@staticmethod
def get_ruler_values(start, stop, max_width, round_to_nearest=True):
if max_width == 0:
return [0]
diff = np.nan_to_num((stop - start) / max_width)
if diff <= 0:
return [0]
decimals = int(np.floor(np.log10(diff)))
if diff > 4 * pow(10, decimals):
step = 5 * pow(10, decimals + 2)
elif diff > 2 * pow(10, decimals):
step = 2 * pow(10, decimals + 2)
elif diff > 1 * pow(10, decimals):
step = 1 * pow(10, decimals + 2)
else:
step = 5 * pow(10, decimals + 1)
round_by = int(-np.floor(np.log10(step)))
r = start % step
if not round_to_nearest:
_range = np.arange(start + step, stop + r, step) - r
start, stop = np.floor(start * 100) / 100, np.ceil(
stop * 100) / 100
return np.round(np.hstack((start, _range, stop)), 2)
return np.round(np.arange(start, stop + r + step, step) - r, round_by)
@staticmethod
def get_points_from_coeffs(current_value, coefficients, possible_values):
if any(np.isnan(possible_values)):
return 0
indices = np.argsort(possible_values)
sorted_values = possible_values[indices]
sorted_coefficients = coefficients[indices]
for i, val in enumerate(sorted_values):
if current_value < val:
break
diff = sorted_values[i] - sorted_values[i - 1]
k = 0 if diff < 1e-6 else (sorted_values[i] - current_value) / \
(sorted_values[i] - sorted_values[i - 1])
return sorted_coefficients[i - 1] * sorted_values[i - 1] * k + \
sorted_coefficients[i] * sorted_values[i] * (1 - k)
class OWBoxPlot(widget.OWWidget):
name = "Box Plot"
description = "Visualize the distribution of feature values in a box plot."
icon = "icons/BoxPlot.svg"
priority = 100
keywords = ["whisker"]
class Inputs:
data = Input("Data", Orange.data.Table)
class Outputs:
selected_data = Output("Selected Data",
Orange.data.Table,
default=True)
annotated_data = Output(ANNOTATED_DATA_SIGNAL_NAME, Orange.data.Table)
class Warning(widget.OWWidget.Warning):
no_vars = widget.Msg(
"Data contains no categorical or numeric variables")
buttons_area_orientation = None
#: Comparison types for continuous variables
CompareNone, CompareMedians, CompareMeans = 0, 1, 2
settingsHandler = DomainContextHandler()
# If this was a list, context handler would try to match its elements to
# variable names!
selection = ContextSetting((), schema_only=True)
attribute = ContextSetting(None)
order_by_importance = Setting(False)
order_grouping_by_importance = Setting(False)
group_var = ContextSetting(None)
show_annotations = Setting(True)
compare = Setting(CompareMeans)
stattest = Setting(0)
sig_threshold = Setting(0.05)
stretched = Setting(True)
show_labels = Setting(True)
sort_freqs = Setting(False)
_sorting_criteria_attrs = {
CompareNone: "",
CompareMedians: "median",
CompareMeans: "mean"
}
_pen_axis_tick = QPen(Qt.white, 5)
_pen_axis = QPen(Qt.darkGray, 3)
_pen_median = QPen(QBrush(QColor(0xff, 0xff, 0x00)), 2)
_pen_paramet = QPen(QBrush(QColor(0x33, 0x00, 0xff)), 2)
_pen_dotted = QPen(QBrush(QColor(0x33, 0x00, 0xff)), 1)
_pen_dotted.setStyle(Qt.DotLine)
_post_line_pen = QPen(Qt.lightGray, 2)
_post_grp_pen = QPen(Qt.lightGray, 4)
for pen in (_pen_paramet, _pen_median, _pen_dotted, _pen_axis,
_pen_axis_tick, _post_line_pen, _post_grp_pen):
pen.setCosmetic(True)
pen.setCapStyle(Qt.RoundCap)
pen.setJoinStyle(Qt.RoundJoin)
_pen_axis_tick.setCapStyle(Qt.FlatCap)
_box_brush = QBrush(QColor(0x33, 0x88, 0xff, 0xc0))
_attr_brush = QBrush(QColor(0x33, 0x00, 0xff))
graph_name = "box_scene"
def __init__(self):
super().__init__()
self._axis_font = QFont()
self._axis_font.setPixelSize(12)
self._label_font = QFont()
self._label_font.setPixelSize(11)
self.dataset = None
self.stats = []
self.dist = self.conts = None
self.posthoc_lines = []
self.label_txts = self.mean_labels = self.boxes = self.labels = \
self.label_txts_all = self.attr_labels = self.order = []
self.scale_x = 1
self.scene_min_x = self.scene_max_x = self.scene_width = 0
self.label_width = 0
self.attrs = VariableListModel()
sorted_model = SortProxyModel(sortRole=Qt.UserRole)
sorted_model.setSourceModel(self.attrs)
sorted_model.sort(0)
box = gui.vBox(self.controlArea, "Variable")
view = self.attr_list = ListViewSearch()
view.setModel(sorted_model)
view.setSelectionMode(view.SingleSelection)
view.selectionModel().selectionChanged.connect(self.attr_changed)
view.setMinimumSize(QSize(30, 30))
# Any other policy than Ignored will let the QListBox's scrollbar
# set the minimal height (see the penultimate paragraph of
# http://doc.qt.io/qt-4.8/qabstractscrollarea.html#addScrollBarWidget)
view.setSizePolicy(QSizePolicy.Expanding, QSizePolicy.Ignored)
box.layout().addWidget(view)
gui.checkBox(box,
self,
"order_by_importance",
"Order by relevance to subgroups",
tooltip="Order by 𝜒² or ANOVA over the subgroups",
callback=self.apply_attr_sorting)
self.group_vars = VariableListModel(placeholder="None")
sorted_model = SortProxyModel(sortRole=Qt.UserRole)
sorted_model.setSourceModel(self.group_vars)
sorted_model.sort(0)
box = gui.vBox(self.controlArea, "Subgroups")
view = self.group_list = ListViewSearch()
view.setModel(sorted_model)
view.selectionModel().selectionChanged.connect(self.grouping_changed)
view.setMinimumSize(QSize(30, 30))
# See the comment above
view.setSizePolicy(QSizePolicy.Expanding, QSizePolicy.Ignored)
box.layout().addWidget(view)
gui.checkBox(box,
self,
"order_grouping_by_importance",
"Order by relevance to variable",
tooltip="Order by 𝜒² or ANOVA over the variable values",
callback=self.apply_group_sorting)
# TODO: move Compare median/mean to grouping box
# The vertical size policy is needed to let only the list views expand
self.display_box = gui.vBox(self.controlArea,
"Display",
sizePolicy=(QSizePolicy.Minimum,
QSizePolicy.Maximum))
gui.checkBox(self.display_box,
self,
"show_annotations",
"Annotate",
callback=self.update_graph)
self.compare_rb = gui.radioButtonsInBox(
self.display_box,
self,
'compare',
btnLabels=["No comparison", "Compare medians", "Compare means"],
callback=self.update_graph)
# The vertical size policy is needed to let only the list views expand
self.stretching_box = box = gui.vBox(self.controlArea,
box="Display",
sizePolicy=(QSizePolicy.Minimum,
QSizePolicy.Fixed))
self.stretching_box.sizeHint = self.display_box.sizeHint
gui.checkBox(box,
self,
'stretched',
"Stretch bars",
callback=self.update_graph,
stateWhenDisabled=False)
gui.checkBox(box,
self,
'show_labels',
"Show box labels",
callback=self.update_graph)
self.sort_cb = gui.checkBox(box,
self,
'sort_freqs',
"Sort by subgroup frequencies",
callback=self.update_graph,
stateWhenDisabled=False)
gui.vBox(self.mainArea)
self.box_scene = QGraphicsScene(self)
self.box_scene.selectionChanged.connect(self.on_selection_changed)
self.box_view = QGraphicsView(self.box_scene)
self.box_view.setRenderHints(QPainter.Antialiasing
| QPainter.TextAntialiasing
| QPainter.SmoothPixmapTransform)
self.box_view.viewport().installEventFilter(self)
self.mainArea.layout().addWidget(self.box_view)
self.stat_test = ""
self.mainArea.setMinimumWidth(300)
self.update_box_visibilities()
def sizeHint(self):
return QSize(900, 500)
def eventFilter(self, obj, event):
if obj is self.box_view.viewport() and \
event.type() == QEvent.Resize:
self.update_graph()
return super().eventFilter(obj, event)
@property
def show_stretched(self):
return self.stretched and self.group_var is not self.attribute
def reset_attrs(self):
domain = self.dataset.domain
self.attrs[:] = [
var for var in chain(domain.class_vars, domain.metas,
domain.attributes)
if var.is_primitive() and not var.attributes.get("hidden", False)
]
def reset_groups(self):
domain = self.dataset.domain
self.group_vars[:] = [None] + [
var for var in chain(domain.class_vars, domain.metas,
domain.attributes)
if var.is_discrete and not var.attributes.get("hidden", False)
]
@Inputs.data
def set_data(self, dataset):
self.closeContext()
self._reset_all_data()
if dataset and not (len(dataset.domain.variables)
or any(var.is_primitive()
for var in dataset.domain.metas)):
self.Warning.no_vars()
dataset = None
self.dataset = dataset
if dataset:
self.reset_attrs()
self.reset_groups()
self._select_default_variables()
self.openContext(self.dataset)
self._set_list_view_selections()
self.compute_box_data()
self.apply_attr_sorting()
self.apply_group_sorting()
self.update_graph()
self.select_box_items()
self.update_box_visibilities()
self.commit()
def _reset_all_data(self):
self.clear_scene()
self.Warning.no_vars.clear()
self.stats = []
self.dist = self.conts = None
self.group_var = None
self.attribute = None
self.stat_test = ""
self.attrs[:] = []
self.group_vars[:] = [None]
self.selection = ()
def _select_default_variables(self):
# visualize first non-class variable, group by class (if present)
domain = self.dataset.domain
if len(self.attrs) > len(domain.class_vars):
self.attribute = self.attrs[len(domain.class_vars)]
elif self.attrs:
self.attribute = self.attrs[0]
if domain.class_var and domain.class_var.is_discrete:
self.group_var = domain.class_var
def _set_list_view_selections(self):
for view, var, callback in ((self.attr_list, self.attribute,
self.attr_changed),
(self.group_list, self.group_var,
self.grouping_changed)):
src_model = view.model().sourceModel()
if var not in src_model:
continue
sel_model = view.selectionModel()
sel_model.selectionChanged.disconnect(callback)
row = src_model.indexOf(var)
index = view.model().index(row, 0)
sel_model.select(index, sel_model.ClearAndSelect)
self._ensure_selection_visible(view)
sel_model.selectionChanged.connect(callback)
def apply_attr_sorting(self):
def compute_score(attr):
# This function and the one in apply_group_sorting are similar, but
# different in too many details, so they are kept as separate
# functions.
# If you discover a bug in this function, check the other one, too.
if attr is group_var:
return 3
if attr.is_continuous:
# One-way ANOVA
col = data.get_column_view(attr)[0].astype(float)
groups = (col[group_col == i] for i in range(n_groups))
groups = (col[~np.isnan(col)] for col in groups)
groups = [group for group in groups if len(group)]
p = f_oneway(*groups)[1] if len(groups) > 1 else 2
else:
p = self._chi_square(group_var, attr)[1]
if math.isnan(p):
return 2
return p
data = self.dataset
if data is None:
return
domain = data.domain
group_var = self.group_var
if self.order_by_importance and group_var is not None:
n_groups = len(group_var.values)
group_col = data.get_column_view(group_var)[0] if \
domain.has_continuous_attributes(
include_class=True, include_metas=True) else None
self._sort_list(self.attrs, self.attr_list, compute_score)
else:
self._sort_list(self.attrs, self.attr_list, None)
def apply_group_sorting(self):
def compute_stat(group):
# This function and the one in apply_attr_sorting are similar, but
# different in too many details, so they are kept as separate
# functions.
# If you discover a bug in this function, check the other one, too.
if group is attr:
return 3
if group is None:
return -1
if attr.is_continuous:
group_col = data.get_column_view(group)[0].astype(float)
groups = (attr_col[group_col == i]
for i in range(len(group.values)))
groups = (col[~np.isnan(col)] for col in groups)
groups = [group for group in groups if len(group)]
p = f_oneway(*groups)[1] if len(groups) > 1 else 2
else:
p = self._chi_square(group, attr)[1]
if math.isnan(p):
return 2
return p
data = self.dataset
if data is None:
return
attr = self.attribute
if self.order_grouping_by_importance:
if attr.is_continuous:
attr_col = data.get_column_view(attr)[0].astype(float)
self._sort_list(self.group_vars, self.group_list, compute_stat)
else:
self._sort_list(self.group_vars, self.group_list, None)
def _sort_list(self, source_model, view, key=None):
if key is None:
c = count()
def key(_): # pylint: disable=function-redefined
return next(c)
for i, attr in enumerate(source_model):
source_model.setData(source_model.index(i), key(attr), Qt.UserRole)
self._ensure_selection_visible(view)
@staticmethod
def _ensure_selection_visible(view):
selection = view.selectedIndexes()
if len(selection) == 1:
view.scrollTo(selection[0])
def _chi_square(self, group_var, attr):
# Chi-square with the given distribution into groups
if not attr.values or not group_var.values:
return 0, 2, 0
observed = np.array(
contingency.get_contingency(self.dataset, group_var, attr))
observed = observed[observed.sum(axis=1) != 0, :]
observed = observed[:, observed.sum(axis=0) != 0]
if min(observed.shape) < 2:
return 0, 2, 0
return chi2_contingency(observed)[:3]
def grouping_changed(self, selected):
if not selected:
return # should never come here
self.group_var = selected.indexes()[0].data(gui.TableVariable)
self._variables_changed(self.apply_attr_sorting)
def attr_changed(self, selected):
if not selected:
return # should never come here
self.attribute = selected.indexes()[0].data(gui.TableVariable)
self._variables_changed(self.apply_group_sorting)
def _variables_changed(self, sorting):
self.selection = ()
self.compute_box_data()
sorting()
self.update_graph()
self.update_box_visibilities()
self.commit()
def update_graph(self):
pending_selection = self.selection
self.box_scene.selectionChanged.disconnect(self.on_selection_changed)
try: # not for exceptions, just to reconnect after all possible paths
self.clear_scene()
if self.dataset is None or self.attribute is None:
return
if self.attribute.is_continuous:
self._display_changed_cont()
else:
self._display_changed_disc()
self.selection = pending_selection
self.draw_stat()
self.select_box_items()
if self.attribute.is_continuous:
heights = 90 if self.show_annotations else 60
self.box_view.centerOn(
self.scene_min_x + self.scene_width / 2,
-30 - len(self.stats) * heights / 2 + 45)
else:
self.box_view.centerOn(self.scene_width / 2,
-30 - len(self.boxes) * 40 / 2 + 45)
finally:
self.box_scene.selectionChanged.connect(self.on_selection_changed)
def select_box_items(self):
selection = set(self.selection)
for box in self.box_scene.items():
if isinstance(box, FilterGraphicsRectItem):
box.setSelected(box.data_range in selection)
def compute_box_data(self):
attr = self.attribute
if not attr:
return
dataset = self.dataset
if dataset is None \
or not attr.is_continuous and not attr.values \
or self.group_var and not self.group_var.values:
self.stats = []
self.dist = self.conts = None
return
if self.group_var:
self.dist = None
self.conts = contingency.get_contingency(dataset, attr,
self.group_var)
missing_val_str = f"missing '{self.group_var.name}'"
group_var_labels = self.group_var.values + ("", )
if self.attribute.is_continuous:
stats, label_texts = [], []
for cont, value in zip(self.conts.array_with_unknowns,
group_var_labels):
if np.sum(cont[1]):
stats.append(BoxData(cont, value))
label_texts.append(value or missing_val_str)
self.stats = stats
self.label_txts_all = label_texts
else:
self.label_txts_all = [
v or missing_val_str for v, c in zip(
group_var_labels, self.conts.array_with_unknowns)
if np.sum(c) > 0
]
else:
self.dist = distribution.get_distribution(dataset, attr)
self.conts = None
if self.attribute.is_continuous:
self.stats = [BoxData(self.dist, None)]
self.label_txts_all = [""]
self.label_txts = [
txts for stat, txts in zip(self.stats, self.label_txts_all)
if stat.n > 0
]
self.stats = [stat for stat in self.stats if stat.n > 0]
def update_box_visibilities(self):
self.controls.stretched.setDisabled(self.group_var is self.attribute)
if not self.attribute:
self.stretching_box.hide()
self.display_box.hide()
elif self.attribute.is_continuous:
self.stretching_box.hide()
self.display_box.show()
self.compare_rb.setEnabled(self.group_var is not None)
else:
self.stretching_box.show()
self.display_box.hide()
self.sort_cb.setEnabled(self.group_var is not None)
def clear_scene(self):
self.box_scene.clear()
self.box_view.viewport().update()
self.attr_labels = []
self.labels = []
self.boxes = []
self.mean_labels = []
self.posthoc_lines = []
def _display_changed_cont(self):
self.mean_labels = [
self.mean_label(stat, self.attribute, lab)
for stat, lab in zip(self.stats, self.label_txts)
]
self.draw_axis()
self.boxes = [self.box_group(stat) for stat in self.stats]
self.labels = [
self.label_group(stat, self.attribute, mean_lab)
for stat, mean_lab in zip(self.stats, self.mean_labels)
]
self.attr_labels = [
QGraphicsSimpleTextItem(lab) for lab in self.label_txts
]
for it in chain(self.labels, self.attr_labels):
self.box_scene.addItem(it)
self.order = list(range(len(self.stats)))
criterion = self._sorting_criteria_attrs[self.compare]
if criterion:
vals = [getattr(stat, criterion) for stat in self.stats]
overmax = max((val for val in vals if val is not None), default=0) \
+ 1
vals = [val if val is not None else overmax for val in vals]
self.order = sorted(self.order, key=vals.__getitem__)
heights = 90 if self.show_annotations else 60
for row, box_index in enumerate(self.order):
y = (-len(self.stats) + row) * heights + 10
for item in self.boxes[box_index]:
self.box_scene.addItem(item)
item.setY(y)
labels = self.labels[box_index]
if self.show_annotations:
labels.show()
labels.setY(y)
else:
labels.hide()
label = self.attr_labels[box_index]
label.setY(y - 15 - label.boundingRect().height())
if self.show_annotations:
label.hide()
else:
stat = self.stats[box_index]
if self.compare == OWBoxPlot.CompareMedians and \
stat.median is not None:
pos = stat.median + 5 / self.scale_x
elif self.compare == OWBoxPlot.CompareMeans or stat.q25 is None:
pos = stat.mean + 5 / self.scale_x
else:
pos = stat.q25
label.setX(pos * self.scale_x)
label.show()
r = QRectF(self.scene_min_x, -30 - len(self.stats) * heights,
self.scene_width,
len(self.stats) * heights + 90)
self.box_scene.setSceneRect(r)
self._compute_tests_cont()
self._show_posthoc()
def _display_changed_disc(self):
self.clear_scene()
self.attr_labels = [
QGraphicsSimpleTextItem(lab) for lab in self.label_txts_all
]
if not self.show_stretched:
if self.group_var:
self.labels = [
QGraphicsTextItem("{}".format(int(sum(cont))))
for cont in self.conts.array_with_unknowns
if np.sum(cont) > 0
]
else:
self.labels = [QGraphicsTextItem(str(int(sum(self.dist))))]
self.order = list(range(len(self.attr_labels)))
self.draw_axis_disc()
if self.group_var:
conts = self.conts.array_with_unknowns
self.boxes = [
self.strudel(cont, val)
for cont, val in zip(conts, self.group_var.values + ("", ))
if np.sum(cont) > 0
]
sums_ = np.sum(conts, axis=1)
sums_ = sums_[sums_ > 0] # only bars with sum > 0 are shown
if self.sort_freqs:
# pylint: disable=invalid-unary-operand-type
self.order = sorted(self.order, key=(-sums_).__getitem__)
else:
conts = self.dist.array_with_unknowns
self.boxes = [self.strudel(conts)]
sums_ = [np.sum(conts)]
for row, box_index in enumerate(self.order):
y = (-len(self.boxes) + row) * 40 + 10
box = self.boxes[box_index]
bars, labels = box[::2], box[1::2]
self.__draw_group_labels(y, box_index)
if not self.show_stretched:
self.__draw_row_counts(y, self.labels[box_index],
sums_[box_index])
if self.show_labels and self.attribute is not self.group_var:
self.__draw_bar_labels(y, bars, labels)
self.__draw_bars(y, bars)
self.box_scene.setSceneRect(-self.label_width - 5,
-30 - len(self.boxes) * 40,
self.scene_width,
len(self.boxes * 40) + 90)
self._compute_tests_disc()
def __draw_group_labels(self, y, row):
"""Draw group labels
Parameters
----------
y: int
vertical offset of bars
row: int
row index
"""
label = self.attr_labels[row]
b = label.boundingRect()
label.setPos(-b.width() - 10, y - b.height() / 2)
self.box_scene.addItem(label)
def __draw_row_counts(self, y, label, row_sum_):
"""Draw row counts
Parameters
----------
y: int
vertical offset of bars
label: QGraphicsSimpleTextItem
Label for group
row_sum_: int
Sum for the group
"""
assert not self.attribute.is_continuous
b = label.boundingRect()
right = self.scale_x * row_sum_
label.setPos(right + 10, y - b.height() / 2)
self.box_scene.addItem(label)
def __draw_bar_labels(self, y, bars, labels):
"""Draw bar labels
Parameters
----------
y: int
vertical offset of bars
bars: List[FilterGraphicsRectItem]
list of bars being drawn
labels: List[QGraphicsTextItem]
list of labels for corresponding bars
"""
for text_item, bar_part in zip(labels, bars):
label = self.Label(text_item.toPlainText())
label.setPos(bar_part.boundingRect().x(),
y - label.boundingRect().height() - 8)
label.setMaxWidth(bar_part.boundingRect().width())
self.box_scene.addItem(label)
def __draw_bars(self, y, bars):
"""Draw bars
Parameters
----------
y: int
vertical offset of bars
bars: List[FilterGraphicsRectItem]
list of bars to draw
"""
for item in bars:
item.setPos(0, y)
self.box_scene.addItem(item)
# noinspection PyPep8Naming
def _compute_tests_cont(self):
# The t-test and ANOVA are implemented here since they efficiently use
# the widget-specific data in self.stats.
# The non-parametric tests can't do this, so we use statistics.tests
# pylint: disable=comparison-with-itself
def stat_ttest():
d1, d2 = self.stats
if d1.n < 2 or d2.n < 2:
return np.nan, np.nan
pooled_var = d1.var / d1.n + d2.var / d2.n
# pylint: disable=comparison-with-itself
if pooled_var == 0 or np.isnan(pooled_var):
return np.nan, np.nan
df = pooled_var ** 2 / \
((d1.var / d1.n) ** 2 / (d1.n - 1) +
(d2.var / d2.n) ** 2 / (d2.n - 1))
t = abs(d1.mean - d2.mean) / math.sqrt(pooled_var)
p = 2 * (1 - scipy.special.stdtr(df, t))
return t, p
# TODO: Check this function
# noinspection PyPep8Naming
def stat_ANOVA():
if any(stat.n == 0 for stat in self.stats):
return np.nan, np.nan
n = sum(stat.n for stat in self.stats)
grand_avg = sum(stat.n * stat.mean for stat in self.stats) / n
var_between = sum(stat.n * (stat.mean - grand_avg)**2
for stat in self.stats)
df_between = len(self.stats) - 1
var_within = sum(stat.n * stat.var for stat in self.stats)
df_within = n - len(self.stats)
if var_within == 0 or df_within == 0 or df_between == 0:
return np.nan, np.nan
F = (var_between / df_between) / (var_within / df_within)
p = 1 - scipy.special.fdtr(df_between, df_within, F)
return F, p
n = len(self.dataset)
if self.compare == OWBoxPlot.CompareNone or len(self.stats) < 2:
t = ""
elif any(s.n <= 1 for s in self.stats):
t = "At least one group has just one instance, " \
"cannot compute significance"
elif len(self.stats) == 2:
if self.compare == OWBoxPlot.CompareMedians:
t = ""
# z, p = tests.wilcoxon_rank_sum(
# self.stats[0].dist, self.stats[1].dist)
# t = "Mann-Whitney's z: %.1f (p=%.3f)" % (z, p)
else:
t, p = stat_ttest()
t = "" if np.isnan(
t) else f"Student's t: {t:.3f} (p={p:.3f}, N={n})"
else:
if self.compare == OWBoxPlot.CompareMedians:
t = ""
# U, p = -1, -1
# t = "Kruskal Wallis's U: %.1f (p=%.3f)" % (U, p)
else:
F, p = stat_ANOVA()
t = "" if np.isnan(F) else f"ANOVA: {F:.3f} (p={p:.3f}, N={n})"
self.stat_test = t
def _compute_tests_disc(self):
if self.group_var is None or self.attribute is None:
self.stat_test = ""
else:
chi, p, dof = self._chi_square(self.group_var, self.attribute)
if np.isnan(p):
self.stat_test = ""
else:
self.stat_test = f"χ²: {chi:.2f} (p={p:.3f}, dof={dof})"
def mean_label(self, stat, attr, val_name):
label = QGraphicsItemGroup()
t = QGraphicsSimpleTextItem(attr.str_val(stat.mean), label)
t.setFont(self._label_font)
bbox = t.boundingRect()
w2, h = bbox.width() / 2, bbox.height()
t.setPos(-w2, -h)
tpm = QGraphicsSimpleTextItem(
" \u00b1 " + "%.*f" % (attr.number_of_decimals + 1, stat.dev),
label)
tpm.setFont(self._label_font)
tpm.setPos(w2, -h)
if val_name:
vnm = QGraphicsSimpleTextItem(val_name + ": ", label)
vnm.setFont(self._label_font)
vnm.setBrush(self._attr_brush)
vb = vnm.boundingRect()
label.min_x = -w2 - vb.width()
vnm.setPos(label.min_x, -h)
else:
label.min_x = -w2
return label
def draw_axis(self):
"""Draw the horizontal axis and sets self.scale_x"""
misssing_stats = not self.stats
stats = self.stats or [BoxData(np.array([[0.], [1.]]), self.attribute)]
mean_labels = self.mean_labels or [
self.mean_label(stats[0], self.attribute, "")
]
bottom = min(stat.a_min for stat in stats)
top = max(stat.a_max for stat in stats)
first_val, step = compute_scale(bottom, top)
while bottom <= first_val:
first_val -= step
bottom = first_val
no_ticks = math.ceil((top - first_val) / step) + 1
top = max(top, first_val + no_ticks * step)
gbottom = min(bottom, min(stat.mean - stat.dev for stat in stats))
gtop = max(top, max(stat.mean + stat.dev for stat in stats))
bv = self.box_view
viewrect = bv.viewport().rect().adjusted(15, 15, -15, -30)
self.scale_x = scale_x = viewrect.width() / (gtop - gbottom)
# In principle we should repeat this until convergence since the new
# scaling is too conservative. (No chance am I doing this.)
mlb = min(stat.mean + mean_lab.min_x / scale_x
for stat, mean_lab in zip(stats, mean_labels))
if mlb < gbottom:
gbottom = mlb
self.scale_x = scale_x = viewrect.width() / (gtop - gbottom)
self.scene_min_x = gbottom * scale_x
self.scene_max_x = gtop * scale_x
self.scene_width = self.scene_max_x - self.scene_min_x
val = first_val
last_text = self.scene_min_x
while True:
l = self.box_scene.addLine(val * scale_x, -1, val * scale_x, 1,
self._pen_axis_tick)
l.setZValue(100)
t = QGraphicsSimpleTextItem(
self.attribute.str_val(val) if not misssing_stats else "?")
t.setFont(self._axis_font)
t.setFlag(QGraphicsItem.ItemIgnoresTransformations)
r = t.boundingRect()
x_start = val * scale_x - r.width() / 2
x_finish = x_start + r.width()
if x_start > last_text + 10 and x_finish < self.scene_max_x:
t.setPos(x_start, 8)
self.box_scene.addItem(t)
last_text = x_finish
if val >= top:
break
val += step
self.box_scene.addLine(bottom * scale_x - 4, 0, top * scale_x + 4, 0,
self._pen_axis)
def draw_stat(self):
if self.stat_test:
label = QGraphicsSimpleTextItem(self.stat_test)
brect = self.box_scene.sceneRect()
label.setPos(brect.center().x() - label.boundingRect().width() / 2,
8 + self._axis_font.pixelSize() * 2)
label.setFlag(QGraphicsItem.ItemIgnoresTransformations)
self.box_scene.addItem(label)
def draw_axis_disc(self):
"""
Draw the horizontal axis and sets self.scale_x for discrete attributes
"""
assert not self.attribute.is_continuous
if self.show_stretched:
if not self.attr_labels:
return
step = steps = 10
else:
if self.group_var:
max_box = max(
float(np.sum(dist))
for dist in self.conts.array_with_unknowns)
else:
max_box = float(np.sum(self.dist.array_with_unknowns))
if max_box == 0:
self.scale_x = 1
return
_, step = compute_scale(0, max_box)
step = int(step) if step > 1 else 1
steps = int(math.ceil(max_box / step))
max_box = step * steps
bv = self.box_view
viewrect = bv.viewport().rect().adjusted(15, 15, -15, -30)
self.scene_width = viewrect.width()
lab_width = max(lab.boundingRect().width() for lab in self.attr_labels)
lab_width = max(lab_width, 40)
lab_width = min(lab_width, self.scene_width / 3)
self.label_width = lab_width
right_offset = 0 # offset for the right label
if not self.show_stretched and self.labels:
if self.group_var:
rows = list(zip(self.conts.array_with_unknowns, self.labels))
else:
rows = [(self.dist, self.labels[0])]
# available space left of the 'group labels'
available = self.scene_width - lab_width - 10
scale_x = (available - right_offset) / max_box
max_right = max(
sum(dist) * scale_x + 10 + lbl.boundingRect().width()
for dist, lbl in rows)
right_offset = max(0, max_right - max_box * scale_x)
self.scale_x = scale_x = \
(self.scene_width - lab_width - 10 - right_offset) / max_box
self.box_scene.addLine(0, 0, max_box * scale_x, 0, self._pen_axis)
for val in range(0, step * steps + 1, step):
l = self.box_scene.addLine(val * scale_x, -1, val * scale_x, 1,
self._pen_axis_tick)
l.setZValue(100)
t = self.box_scene.addSimpleText(str(val), self._axis_font)
t.setPos(val * scale_x - t.boundingRect().width() / 2, 8)
if self.show_stretched:
self.scale_x *= 100
def label_group(self, stat, attr, mean_lab):
def centered_text(val, pos):
t = QGraphicsSimpleTextItem(attr.str_val(val), labels)
t.setFont(self._label_font)
bbox = t.boundingRect()
t.setPos(pos - bbox.width() / 2, 22)
return t
def line(x, down=1):
QGraphicsLineItem(x, 12 * down, x, 20 * down, labels)
def move_label(label, frm, to):
label.setX(to)
to += t_box.width() / 2
path = QPainterPath()
path.lineTo(0, 4)
path.lineTo(to - frm, 4)
path.lineTo(to - frm, 8)
p = QGraphicsPathItem(path)
p.setPos(frm, 12)
labels.addToGroup(p)
labels = QGraphicsItemGroup()
labels.addToGroup(mean_lab)
m = stat.mean * self.scale_x
mean_lab.setPos(m, -22)
line(m, -1)
if stat.median is not None:
msc = stat.median * self.scale_x
med_t = centered_text(stat.median, msc)
med_box_width2 = med_t.boundingRect().width() / 2
line(msc)
if stat.q25 is not None:
x = stat.q25 * self.scale_x
t = centered_text(stat.q25, x)
t_box = t.boundingRect()
med_left = msc - med_box_width2
if x + t_box.width() / 2 >= med_left - 5:
move_label(t, x, med_left - t_box.width() - 5)
else:
line(x)
if stat.q75 is not None:
x = stat.q75 * self.scale_x
t = centered_text(stat.q75, x)
t_box = t.boundingRect()
med_right = msc + med_box_width2
if x - t_box.width() / 2 <= med_right + 5:
move_label(t, x, med_right + 5)
else:
line(x)
return labels
def box_group(self, stat, height=20):
def line(x0, y0, x1, y1, *args):
return QGraphicsLineItem(x0 * scale_x, y0, x1 * scale_x, y1, *args)
scale_x = self.scale_x
box = []
whisker1 = line(stat.a_min, -1.5, stat.a_min, 1.5)
whisker2 = line(stat.a_max, -1.5, stat.a_max, 1.5)
vert_line = line(stat.a_min, 0, stat.a_max, 0)
mean_line = line(stat.mean, -height / 3, stat.mean, height / 3)
for it in (whisker1, whisker2, mean_line):
it.setPen(self._pen_paramet)
vert_line.setPen(self._pen_dotted)
var_line = line(stat.mean - stat.dev, 0, stat.mean + stat.dev, 0)
var_line.setPen(self._pen_paramet)
box.extend([whisker1, whisker2, vert_line, mean_line, var_line])
if stat.q25 is not None or stat.q75 is not None:
# if any of them is None it means that its value is equal to median
box_from = stat.q25 or stat.median
box_to = stat.q75 or stat.median
mbox = FilterGraphicsRectItem(stat.data_range, box_from * scale_x,
-height / 2,
(box_to - box_from) * scale_x,
height)
mbox.setBrush(self._box_brush)
mbox.setPen(QPen(Qt.NoPen))
mbox.setZValue(-200)
box.append(mbox)
if stat.median is not None:
median_line = line(stat.median, -height / 2, stat.median,
height / 2)
median_line.setPen(self._pen_median)
median_line.setZValue(-150)
box.append(median_line)
return box
def strudel(self, dist, group_val=None):
attr = self.attribute
ss = np.sum(dist)
box = []
if ss < 1e-6:
cond = DiscDataRange(None, group_val)
box.append(FilterGraphicsRectItem(cond, 0, -10, 1, 10))
cum = 0
missing_val_str = f"missing '{attr.name}'"
values = attr.values + ("", )
colors = attr.palette.qcolors_w_nan
total = sum(dist)
for freq, value, color in zip(dist, values, colors):
if freq < 1e-6:
continue
v = freq
if self.show_stretched:
v /= ss
v *= self.scale_x
cond = DiscDataRange(value, group_val)
rect = FilterGraphicsRectItem(cond, cum + 1, -6, v - 2, 12)
rect.setBrush(QBrush(color))
rect.setPen(QPen(Qt.NoPen))
value = value or missing_val_str
if self.show_stretched:
tooltip = f"{value}: {100 * freq / total:.2f}%"
else:
tooltip = f"{value}: ({int(freq)})"
rect.setToolTip(tooltip)
text = QGraphicsTextItem(value)
box.append(rect)
box.append(text)
cum += v
return box
def on_selection_changed(self):
self.selection = tuple(item.data_range
for item in self.box_scene.selectedItems()
if item.data_range)
self.commit()
def commit(self):
conditions = self._gather_conditions()
if conditions:
selected = Values(conditions, conjunction=False)(self.dataset)
selection = np.in1d(self.dataset.ids,
selected.ids,
assume_unique=True).nonzero()[0]
else:
selected, selection = None, []
self.Outputs.selected_data.send(selected)
self.Outputs.annotated_data.send(
create_annotated_table(self.dataset, selection))
def _gather_conditions(self):
conditions = []
attr = self.attribute
group_attr = self.group_var
for data_range in self.selection:
if attr.is_discrete:
# If some value was removed from the data (in case settings are
# loaded from a scheme), do not include the corresponding
# filter; this is appropriate since data with such value does
# not exist anyway
if not data_range.value:
condition = IsDefined([attr], negate=True)
elif data_range.value not in attr.values:
continue
else:
condition = FilterDiscrete(attr, [data_range.value])
else:
condition = FilterContinuous(attr, FilterContinuous.Between,
data_range.low, data_range.high)
if data_range.group_value:
if not data_range.group_value:
grp_filter = IsDefined([group_attr], negate=True)
elif data_range.group_value not in group_attr.values:
continue
else:
grp_filter = FilterDiscrete(group_attr,
[data_range.group_value])
condition = Values([condition, grp_filter], conjunction=True)
conditions.append(condition)
return conditions
def _show_posthoc(self):
def line(y0, y1):
it = self.box_scene.addLine(x, y0, x, y1, self._post_line_pen)
it.setZValue(-100)
self.posthoc_lines.append(it)
while self.posthoc_lines:
self.box_scene.removeItem(self.posthoc_lines.pop())
if self.compare == OWBoxPlot.CompareNone or len(self.stats) < 2:
return
if self.compare == OWBoxPlot.CompareMedians:
crit_line = "median"
else:
crit_line = "mean"
xs = []
height = 90 if self.show_annotations else 60
y_up = -len(self.stats) * height + 10
for pos, box_index in enumerate(self.order):
stat = self.stats[box_index]
x = getattr(stat, crit_line)
if x is None:
continue
x *= self.scale_x
xs.append(x * self.scale_x)
by = y_up + pos * height
line(by + 12, 0)
used_to = []
last_to = to = 0
for frm, frm_x in enumerate(xs[:-1]):
for to in range(frm + 1, len(xs)):
if xs[to] - frm_x > 1.5:
to -= 1
break
if to in (last_to, frm):
continue
for rowi, used in enumerate(used_to):
if used < frm:
used_to[rowi] = to
break
else:
rowi = len(used_to)
used_to.append(to)
y = -6 - rowi * 6
it = self.box_scene.addLine(frm_x - 2, y, xs[to] + 2, y,
self._post_grp_pen)
self.posthoc_lines.append(it)
last_to = to
def get_widget_name_extension(self):
return self.attribute.name if self.attribute else None
def send_report(self):
self.report_plot()
text = ""
if self.attribute:
text += "Box plot for attribute '{}' ".format(self.attribute.name)
if self.group_var:
text += "grouped by '{}'".format(self.group_var.name)
if text:
self.report_caption(text)
class Label(QGraphicsSimpleTextItem):
"""Boxplot Label with settable maxWidth"""
# Minimum width to display label text
MIN_LABEL_WIDTH = 25
# padding bellow the text
PADDING = 3
__max_width = None
def maxWidth(self):
return self.__max_width
def setMaxWidth(self, max_width):
self.__max_width = max_width
def paint(self, painter, option, widget):
"""Overrides QGraphicsSimpleTextItem.paint
If label text is too long, it is elided
to fit into the allowed region
"""
if self.__max_width is None:
width = option.rect.width()
else:
width = self.__max_width
if width < self.MIN_LABEL_WIDTH:
# if space is too narrow, no label
return
fm = painter.fontMetrics()
text = fm.elidedText(self.text(), Qt.ElideRight, width)
painter.drawText(
option.rect.x(),
option.rect.y() + self.boundingRect().height() - self.PADDING,
text)
def data(self, index, role):
# type: (QModelIndex, Qt.ItemDataRole) -> Any
# Text formatting for various data simply requires a lot of branches.
# This is much better than overengineering various formatters...
# pylint: disable=too-many-branches
if not index.isValid():
return None
row, column = self.mapToSourceRows(index.row()), index.column()
# Make sure we're not out of range
if not 0 <= row <= self.n_attributes:
return QVariant()
attribute = self.variables[row]
if role == Qt.BackgroundRole:
if attribute in self.domain.attributes:
return self.COLOR_FOR_ROLE[self.ATTRIBUTE]
elif attribute in self.domain.metas:
return self.COLOR_FOR_ROLE[self.META]
elif attribute in self.domain.class_vars:
return self.COLOR_FOR_ROLE[self.CLASS_VAR]
elif role == Qt.TextAlignmentRole:
if column == self.Columns.NAME:
return Qt.AlignLeft | Qt.AlignVCenter
return Qt.AlignRight | Qt.AlignVCenter
output = None
if column == self.Columns.ICON:
if role == Qt.DecorationRole:
return gui.attributeIconDict[attribute]
elif column == self.Columns.NAME:
if role == Qt.DisplayRole:
output = attribute.name
elif column == self.Columns.DISTRIBUTION:
if role == Qt.DisplayRole:
if isinstance(attribute, (DiscreteVariable, ContinuousVariable)):
if row not in self.__distributions_cache:
scene = QGraphicsScene(parent=self)
histogram = Histogram(
data=self.table,
variable=attribute,
color_attribute=self.target_var,
border=(0, 0, 2, 0),
border_color='#ccc',
)
scene.addItem(histogram)
self.__distributions_cache[row] = scene
return self.__distributions_cache[row]
elif column == self.Columns.CENTER:
if role == Qt.DisplayRole:
if isinstance(attribute, DiscreteVariable):
output = self._center[row]
if not np.isnan(output):
output = attribute.str_val(self._center[row])
elif isinstance(attribute, TimeVariable):
output = attribute.str_val(self._center[row])
else:
output = self._center[row]
elif column == self.Columns.DISPERSION:
if role == Qt.DisplayRole:
if isinstance(attribute, TimeVariable):
output = format_time_diff(self._min[row], self._max[row])
else:
output = self._dispersion[row]
elif column == self.Columns.MIN:
if role == Qt.DisplayRole:
if isinstance(attribute, DiscreteVariable):
if attribute.ordered:
output = attribute.str_val(self._min[row])
elif isinstance(attribute, TimeVariable):
output = attribute.str_val(self._min[row])
else:
output = self._min[row]
elif column == self.Columns.MAX:
if role == Qt.DisplayRole:
if isinstance(attribute, DiscreteVariable):
if attribute.ordered:
output = attribute.str_val(self._max[row])
elif isinstance(attribute, TimeVariable):
output = attribute.str_val(self._max[row])
else:
output = self._max[row]
elif column == self.Columns.MISSING:
if role == Qt.DisplayRole:
output = '%d (%d%%)' % (
self._missing[row],
100 * self._missing[row] / self.n_instances
)
# Consistently format the text inside the table cells
# The easiest way to check for NaN is to compare with itself
if output != output: # pylint: disable=comparison-with-itself
output = ''
# Format ∞ properly
elif output in (np.inf, -np.inf):
output = '%s∞' % ['', '-'][output < 0]
elif isinstance(output, int):
output = locale.format('%d', output, grouping=True)
elif isinstance(output, float):
output = locale.format('%.2f', output, grouping=True)
return output
class OWVennDiagram(widget.OWWidget):
name = "Venn Diagram"
description = "A graphical visualization of the overlap of data instances " "from a collection of input data sets."
icon = "icons/VennDiagram.svg"
priority = 280
inputs = [("Data", Orange.data.Table, "setData", widget.Multiple)]
outputs = [("Selected Data", Orange.data.Table)]
# Selected disjoint subset indices
selection = settings.Setting([])
#: Stored input set hints
#: {(index, inputname, attributes): (selectedattrname, itemsettitle)}
#: The 'selectedattrname' can be None
inputhints = settings.Setting({})
#: Use identifier columns for instance matching
useidentifiers = settings.Setting(True)
#: Output unique items (one output row for every unique instance `key`)
#: or preserve all duplicates in the output.
output_duplicates = settings.Setting(False)
autocommit = settings.Setting(True)
graph_name = "scene"
def __init__(self):
super().__init__()
# Diagram update is in progress
self._updating = False
# Input update is in progress
self._inputUpdate = False
# All input tables have the same domain.
self.samedomain = True
# Input datasets in the order they were 'connected'.
self.data = OrderedDict()
# Extracted input item sets in the order they were 'connected'
self.itemsets = OrderedDict()
# GUI
box = gui.vBox(self.controlArea, "Info")
self.info = gui.widgetLabel(box, "No data on input.\n")
self.identifiersBox = gui.radioButtonsInBox(
self.controlArea,
self,
"useidentifiers",
[],
box="Data Instance Identifiers",
callback=self._on_useidentifiersChanged,
)
self.useequalityButton = gui.appendRadioButton(self.identifiersBox, "Use instance equality")
self.useidentifiersButton = rb = gui.appendRadioButton(self.identifiersBox, "Use identifiers")
self.inputsBox = gui.indentedBox(self.identifiersBox, sep=gui.checkButtonOffsetHint(rb))
self.inputsBox.setEnabled(bool(self.useidentifiers))
for i in range(5):
box = gui.vBox(self.inputsBox, "Data set #%i" % (i + 1), addSpace=False)
box.setFlat(True)
model = itemmodels.VariableListModel(parent=self)
cb = QComboBox(minimumContentsLength=12, sizeAdjustPolicy=QComboBox.AdjustToMinimumContentsLengthWithIcon)
cb.setModel(model)
cb.activated[int].connect(self._on_inputAttrActivated)
box.setEnabled(False)
# Store the combo in the box for later use.
box.combo_box = cb
box.layout().addWidget(cb)
gui.rubber(self.controlArea)
box = gui.vBox(self.controlArea, "Output")
gui.checkBox(box, self, "output_duplicates", "Output duplicates", callback=lambda: self.commit())
gui.auto_commit(box, self, "autocommit", "Send Selection", "Send Automatically", box=False)
# Main area view
self.scene = QGraphicsScene()
self.view = QGraphicsView(self.scene)
self.view.setRenderHint(QPainter.Antialiasing)
self.view.setBackgroundRole(QPalette.Window)
self.view.setFrameStyle(QGraphicsView.StyledPanel)
self.mainArea.layout().addWidget(self.view)
self.vennwidget = VennDiagram()
self.vennwidget.resize(400, 400)
self.vennwidget.itemTextEdited.connect(self._on_itemTextEdited)
self.scene.selectionChanged.connect(self._on_selectionChanged)
self.scene.addItem(self.vennwidget)
self.resize(self.controlArea.sizeHint().width() + 550, max(self.controlArea.sizeHint().height(), 550))
self._queue = []
@check_sql_input
def setData(self, data, key=None):
self.error()
if not self._inputUpdate:
# Store hints only on the first setData call.
self._storeHints()
self._inputUpdate = True
if key in self.data:
if data is None:
# Remove the input
self._remove(key)
else:
# Update existing item
self._update(key, data)
elif data is not None:
# TODO: Allow setting more them 5 inputs and let the user
# select the 5 to display.
if len(self.data) == 5:
self.error("Venn diagram accepts at most five data sets.")
return
# Add a new input
self._add(key, data)
def handleNewSignals(self):
self._inputUpdate = False
# Check if all inputs are from the same domain.
domains = [input.table.domain for input in self.data.values()]
samedomain = all(domain_eq(d1, d2) for d1, d2 in pairwise(domains))
self.samedomain = samedomain
has_identifiers = all(source_attributes(input.table.domain) for input in self.data.values())
has_any_identifiers = any(source_attributes(input.table.domain) for input in self.data.values())
self.useequalityButton.setEnabled(samedomain)
self.useidentifiersButton.setEnabled(has_any_identifiers or len(self.data) == 0)
self.inputsBox.setEnabled(has_any_identifiers)
if not samedomain and has_any_identifiers and not self.useidentifiers:
self.useidentifiers = 1
elif samedomain and not has_identifiers:
self.useidentifiers = 0
incremental = all(inc for _, inc in self._queue)
if incremental:
# Only received updated data on existing link.
self._updateItemsets()
else:
# Links were removed and/or added.
self._createItemsets()
self._restoreHints()
self._updateItemsets()
del self._queue[:]
self._createDiagram()
if self.data:
self.info.setText("{} data sets on input.\n".format(len(self.data)))
else:
self.info.setText("No data on input\n")
self._updateInfo()
super().handleNewSignals()
def _invalidate(self, keys=None, incremental=True):
"""
Invalidate input for a list of input keys.
"""
if keys is None:
keys = list(self.data.keys())
self._queue.extend((key, incremental) for key in keys)
def itemsetAttr(self, key):
index = list(self.data.keys()).index(key)
_, combo = self._controlAtIndex(index)
model = combo.model()
attr_index = combo.currentIndex()
if attr_index >= 0:
return model[attr_index]
else:
return None
def _controlAtIndex(self, index):
group_box = self.inputsBox.layout().itemAt(index).widget()
combo = group_box.combo_box
return group_box, combo
def _setAttributes(self, index, attrs):
box, combo = self._controlAtIndex(index)
model = combo.model()
if attrs is None:
model[:] = []
box.setEnabled(False)
else:
if model[:] != attrs:
model[:] = attrs
box.setEnabled(True)
def _add(self, key, table):
name = table.name
index = len(self.data)
attrs = source_attributes(table.domain)
self.data[key] = _InputData(key, name, table)
self._setAttributes(index, attrs)
self._invalidate([key], incremental=False)
item = self.inputsBox.layout().itemAt(index)
box = item.widget()
box.setTitle("Data set: {}".format(name))
def _remove(self, key):
index = list(self.data.keys()).index(key)
# Clear possible warnings.
self.warning()
self._setAttributes(index, None)
del self.data[key]
layout = self.inputsBox.layout()
item = layout.takeAt(index)
layout.addItem(item)
inputs = list(self.data.values())
for i in range(5):
box, _ = self._controlAtIndex(i)
if i < len(inputs):
title = "Data set: {}".format(inputs[i].name)
else:
title = "Data set #{}".format(i + 1)
box.setTitle(title)
self._invalidate([key], incremental=False)
def _update(self, key, table):
name = table.name
index = list(self.data.keys()).index(key)
attrs = source_attributes(table.domain)
self.data[key] = self.data[key]._replace(name=name, table=table)
self._setAttributes(index, attrs)
self._invalidate([key])
item = self.inputsBox.layout().itemAt(index)
box = item.widget()
box.setTitle("Data set: {}".format(name))
def _itemsForInput(self, key):
useidentifiers = self.useidentifiers or not self.samedomain
def items_by_key(key, input):
attr = self.itemsetAttr(key)
if attr is not None:
return [str(inst[attr]) for inst in input.table if not numpy.isnan(inst[attr])]
else:
return []
def items_by_eq(key, input):
return list(map(ComparableInstance, input.table))
input = self.data[key]
if useidentifiers:
items = items_by_key(key, input)
else:
items = items_by_eq(key, input)
return items
def _updateItemsets(self):
assert list(self.data.keys()) == list(self.itemsets.keys())
for key, input in list(self.data.items()):
items = self._itemsForInput(key)
item = self.itemsets[key]
item = item._replace(items=items)
name = input.name
if item.name != name:
item = item._replace(name=name, title=name)
self.itemsets[key] = item
def _createItemsets(self):
olditemsets = dict(self.itemsets)
self.itemsets.clear()
for key, input in self.data.items():
items = self._itemsForInput(key)
name = input.name
if key in olditemsets and olditemsets[key].name == name:
# Reuse the title (which might have been changed by the user)
title = olditemsets[key].title
else:
title = name
itemset = _ItemSet(key=key, name=name, title=title, items=items)
self.itemsets[key] = itemset
def _storeHints(self):
if self.data:
self.inputhints.clear()
for i, (key, input) in enumerate(self.data.items()):
attrs = source_attributes(input.table.domain)
attrs = tuple(attr.name for attr in attrs)
selected = self.itemsetAttr(key)
if selected is not None:
attr_name = selected.name
else:
attr_name = None
itemset = self.itemsets[key]
self.inputhints[(i, input.name, attrs)] = (attr_name, itemset.title)
def _restoreHints(self):
settings = []
for i, (key, input) in enumerate(self.data.items()):
attrs = source_attributes(input.table.domain)
attrs = tuple(attr.name for attr in attrs)
hint = self.inputhints.get((i, input.name, attrs), None)
if hint is not None:
attr, name = hint
attr_ind = attrs.index(attr) if attr is not None else -1
settings.append((attr_ind, name))
else:
return
# all inputs match the stored hints
for i, key in enumerate(self.itemsets):
attr, itemtitle = settings[i]
self.itemsets[key] = self.itemsets[key]._replace(title=itemtitle)
_, cb = self._controlAtIndex(i)
cb.setCurrentIndex(attr)
def _createDiagram(self):
self._updating = True
oldselection = list(self.selection)
self.vennwidget.clear()
n = len(self.itemsets)
self.disjoint = disjoint(set(s.items) for s in self.itemsets.values())
vennitems = []
colors = colorpalette.ColorPaletteHSV(n)
for i, (key, item) in enumerate(self.itemsets.items()):
count = len(set(item.items))
count_all = len(item.items)
if count != count_all:
fmt = "{} <i>(all: {})</i>"
else:
fmt = "{}"
counts = fmt.format(count, count_all)
gr = VennSetItem(text=item.title, informativeText=counts)
color = colors[i]
color.setAlpha(100)
gr.setBrush(QBrush(color))
gr.setPen(QPen(Qt.NoPen))
vennitems.append(gr)
self.vennwidget.setItems(vennitems)
for i, area in enumerate(self.vennwidget.vennareas()):
area_items = list(map(str, list(self.disjoint[i])))
if i:
area.setText("{0}".format(len(area_items)))
label = disjoint_set_label(i, n, simplify=False)
head = "<h4>|{}| = {}</h4>".format(label, len(area_items))
if len(area_items) > 32:
items_str = ", ".join(map(escape, area_items[:32]))
hidden = len(area_items) - 32
tooltip = "{}<span>{}, ...</br>({} items not shown)<span>".format(head, items_str, hidden)
elif area_items:
tooltip = "{}<span>{}</span>".format(head, ", ".join(map(escape, area_items)))
else:
tooltip = head
area.setToolTip(tooltip)
area.setPen(QPen(QColor(10, 10, 10, 200), 1.5))
area.setFlag(QGraphicsPathItem.ItemIsSelectable, True)
area.setSelected(i in oldselection)
self._updating = False
self._on_selectionChanged()
def _updateInfo(self):
# Clear all warnings
self.warning()
if not len(self.data):
self.info.setText("No data on input\n")
else:
self.info.setText("{0} data sets on input\n".format(len(self.data)))
if self.useidentifiers:
no_idx = [
"#{}".format(i + 1)
for i, key in enumerate(self.data)
if not source_attributes(self.data[key].table.domain)
]
if len(no_idx) == 1:
self.warning("Data set {} has no suitable identifiers.".format(no_idx[0]))
elif len(no_idx) > 1:
self.warning(
"Data sets {} and {} have no suitable identifiers.".format(", ".join(no_idx[:-1]), no_idx[-1])
)
def _on_selectionChanged(self):
if self._updating:
return
areas = self.vennwidget.vennareas()
indices = [i for i, area in enumerate(areas) if area.isSelected()]
self.selection = indices
self.invalidateOutput()
def _on_useidentifiersChanged(self):
self.inputsBox.setEnabled(self.useidentifiers == 1)
# Invalidate all itemsets
self._invalidate()
self._updateItemsets()
self._createDiagram()
self._updateInfo()
def _on_inputAttrActivated(self, attr_index):
combo = self.sender()
# Find the input index to which the combo box belongs
# (they are reordered when removing inputs).
index = None
inputs = list(self.data.items())
for i in range(len(inputs)):
_, c = self._controlAtIndex(i)
if c is combo:
index = i
break
assert index is not None
key, _ = inputs[index]
self._invalidate([key])
self._updateItemsets()
self._createDiagram()
def _on_itemTextEdited(self, index, text):
text = str(text)
key = list(self.itemsets.keys())[index]
self.itemsets[key] = self.itemsets[key]._replace(title=text)
def invalidateOutput(self):
self.commit()
def commit(self):
selected_subsets = []
selected_items = reduce(set.union, [self.disjoint[index] for index in self.selection], set())
def match(val):
if numpy.isnan(val):
return False
else:
return str(val) in selected_items
source_var = Orange.data.StringVariable("source")
item_id_var = Orange.data.StringVariable("item_id")
names = [itemset.title.strip() for itemset in self.itemsets.values()]
names = uniquify(names)
for i, (key, input) in enumerate(self.data.items()):
if self.useidentifiers:
attr = self.itemsetAttr(key)
if attr is not None:
mask = list(map(match, (inst[attr] for inst in input.table)))
else:
mask = [False] * len(input.table)
def instance_key(inst):
return str(inst[attr])
else:
mask = [ComparableInstance(inst) in selected_items for inst in input.table]
_map = {item: str(i) for i, item in enumerate(selected_items)}
def instance_key(inst):
return _map[ComparableInstance(inst)]
mask = numpy.array(mask, dtype=bool)
subset = input.table[mask]
if len(subset) == 0:
continue
# add columns with source table id and set id
if not self.output_duplicates:
id_column = numpy.array([[instance_key(inst)] for inst in subset], dtype=object)
source_names = numpy.array([[names[i]]] * len(subset), dtype=object)
subset = append_column(subset, "M", source_var, source_names)
subset = append_column(subset, "M", item_id_var, id_column)
selected_subsets.append(subset)
if selected_subsets and not self.output_duplicates:
data = table_concat(selected_subsets)
# Get all variables which are not constant between the same
# item set
varying = varying_between(data, [item_id_var])
if source_var in varying:
varying.remove(source_var)
data = reshape_wide(data, varying, [item_id_var], [source_var])
# remove the temporary item set id column
data = drop_columns(data, [item_id_var])
elif selected_subsets:
data = table_concat(selected_subsets)
else:
data = None
self.send("Selected Data", data)
def getSettings(self, *args, **kwargs):
self._storeHints()
return super().getSettings(self, *args, **kwargs)
def send_report(self):
self.report_plot()
class OWPieChart(widget.OWWidget):
name = "Pie Chart"
description = "Make fun of Pie Charts."
icon = "icons/PieChart.svg"
priority = 100
class Inputs:
data = Input("Data", Orange.data.Table)
settingsHandler = DomainContextHandler()
attribute = ContextSetting(None)
split_var = ContextSetting(None)
explode = Setting(False)
graph_name = "scene"
def __init__(self):
super().__init__()
self.dataset = None
self.attrs = DomainModel(
valid_types=Orange.data.DiscreteVariable, separators=False)
cb = gui.comboBox(
self.controlArea, self, "attribute", box=True,
model=self.attrs, callback=self.update_scene, contentsLength=12)
grid = QGridLayout()
self.legend = gui.widgetBox(gui.indentedBox(cb.box), orientation=grid)
grid.setColumnStretch(1, 1)
grid.setHorizontalSpacing(6)
self.legend_items = []
self.split_vars = DomainModel(
valid_types=Orange.data.DiscreteVariable, separators=False,
placeholder="None", )
gui.comboBox(
self.controlArea, self, "split_var", box="Split by",
model=self.split_vars, callback=self.update_scene)
gui.checkBox(
self.controlArea, self, "explode", "Explode pies", box=True,
callback=self.update_scene)
gui.rubber(self.controlArea)
gui.widgetLabel(
gui.hBox(self.controlArea, box=True),
"The aim of this widget is to\n"
"demonstrate that pie charts are\n"
"a terrible visualization. Please\n"
"don't use it for any other purpose.")
self.scene = QGraphicsScene()
self.view = QGraphicsView(self.scene)
self.view.setRenderHints(
QPainter.Antialiasing | QPainter.TextAntialiasing |
QPainter.SmoothPixmapTransform)
self.mainArea.layout().addWidget(self.view)
self.mainArea.setMinimumWidth(600)
def sizeHint(self):
return QSize(200, 150) # Horizontal size is regulated by mainArea
@Inputs.data
def set_data(self, dataset):
if dataset is not None and (
not bool(dataset) or not len(dataset.domain)):
dataset = None
self.closeContext()
self.dataset = dataset
self.attribute = None
self.split_var = None
domain = dataset.domain if dataset is not None else None
self.attrs.set_domain(domain)
self.split_vars.set_domain(domain)
if dataset is not None:
self.select_default_variables(domain)
self.openContext(self.dataset)
self.update_scene()
def select_default_variables(self, domain):
if len(self.attrs) > len(domain.class_vars):
first_attr = self.split_vars[len(domain.class_vars)]
else:
first_attr = None
if len(self.attrs):
self.attribute, self.split_var = self.attrs[0], first_attr
else:
self.attribute, self.split_var = self.split_var, None
def update_scene(self):
self.scene.clear()
if self.dataset is None or self.attribute is None:
return
dists, labels = self.compute_box_data()
colors = self.attribute.colors
for x, (dist, label) in enumerate(zip(dists, labels)):
self.pie_chart(SCALE * x, 0, 0.8 * SCALE, dist, colors)
self.pie_label(SCALE * x, 0, label)
self.update_legend(
[QColor(*col) for col in colors], self.attribute.values)
self.view.centerOn(SCALE * len(dists) / 2, 0)
def update_legend(self, colors, labels):
layout = self.legend.layout()
while self.legend_items:
w = self.legend_items.pop()
layout.removeWidget(w)
w.deleteLater()
for row, (color, label) in enumerate(zip(colors, labels)):
icon = QLabel()
p = QPixmap(12, 12)
p.fill(color)
icon.setPixmap(p)
label = QLabel(label)
layout.addWidget(icon, row, 0)
layout.addWidget(label, row, 1, alignment=Qt.AlignLeft)
self.legend_items += (icon, label)
def pie_chart(self, x, y, r, dist, colors):
start_angle = 0
dist = np.asarray(dist)
spans = dist / (float(np.sum(dist)) or 1) * 360 * 16
for span, color in zip(spans, colors):
if not span:
continue
if self.explode:
mid_ang = (start_angle + span / 2) / 360 / 16 * 2 * pi
dx = r / 30 * cos(mid_ang)
dy = r / 30 * sin(mid_ang)
else:
dx = dy = 0
ellipse = QGraphicsEllipseItem(x - r / 2 + dx, y - r / 2 - dy, r, r)
if len(spans) > 1:
ellipse.setStartAngle(start_angle)
ellipse.setSpanAngle(span)
ellipse.setBrush(QColor(*color))
self.scene.addItem(ellipse)
start_angle += span
def pie_label(self, x, y, label):
if not label:
return
text = QGraphicsSimpleTextItem(label)
for cut in range(1, len(label)):
if text.boundingRect().width() < 0.95 * SCALE:
break
text = QGraphicsSimpleTextItem(label[:-cut] + "...")
text.setPos(x - text.boundingRect().width() / 2, y + 0.5 * SCALE)
self.scene.addItem(text)
def compute_box_data(self):
if self.split_var:
return (
contingency.get_contingency(
self.dataset, self.attribute, self.split_var),
self.split_var.values)
else:
return [
distribution.get_distribution(
self.dataset, self.attribute)], [""]
def send_report(self):
self.report_plot()
text = ""
if self.attribute is not None:
text += "Box plot for '{}' ".format(self.attribute.name)
if self.split_var is not None:
text += "split by '{}'".format(self.split_var.name)
if text:
self.report_caption(text)
class OWPythagoreanForest(OWWidget):
name = 'Pythagorean Forest'
description = 'Pythagorean forest for visualising random forests.'
icon = 'icons/PythagoreanForest.svg'
priority = 1001
inputs = [('Random forest', RandomForestModel, 'set_rf')]
outputs = [('Tree', TreeModel)]
# Enable the save as feature
graph_name = 'scene'
# Settings
depth_limit = settings.ContextSetting(10)
target_class_index = settings.ContextSetting(0)
size_calc_idx = settings.Setting(0)
size_log_scale = settings.Setting(2)
zoom = settings.Setting(50)
selected_tree_index = settings.ContextSetting(-1)
CLASSIFICATION, REGRESSION = range(2)
def __init__(self):
super().__init__()
# Instance variables
self.forest_type = self.CLASSIFICATION
self.model = None
self.forest_adapter = None
self.dataset = None
self.clf_dataset = None
# We need to store refernces to the trees and grid items
self.grid_items, self.ptrees = [], []
self.color_palette = None
# Different methods to calculate the size of squares
self.SIZE_CALCULATION = [
('Normal', lambda x: x),
('Square root', lambda x: sqrt(x)),
('Logarithmic', lambda x: log(x * self.size_log_scale)),
]
self.REGRESSION_COLOR_CALC = [
('None', lambda _, __: QColor(255, 255, 255)),
('Class mean', self._color_class_mean),
('Standard deviation', self._color_stddev),
]
# CONTROL AREA
# Tree info area
box_info = gui.widgetBox(self.controlArea, 'Forest')
self.ui_info = gui.widgetLabel(box_info, label='')
# Display controls area
box_display = gui.widgetBox(self.controlArea, 'Display')
self.ui_depth_slider = gui.hSlider(
box_display, self, 'depth_limit', label='Depth', ticks=False,
callback=self.max_depth_changed)
self.ui_target_class_combo = gui.comboBox(
box_display, self, 'target_class_index', label='Target class',
orientation=Qt.Horizontal, items=[], contentsLength=8,
callback=self.target_colors_changed)
self.ui_size_calc_combo = gui.comboBox(
box_display, self, 'size_calc_idx', label='Size',
orientation=Qt.Horizontal,
items=list(zip(*self.SIZE_CALCULATION))[0], contentsLength=8,
callback=self.size_calc_changed)
self.ui_zoom_slider = gui.hSlider(
box_display, self, 'zoom', label='Zoom', ticks=False, minValue=20,
maxValue=150, callback=self.zoom_changed, createLabel=False)
# Stretch to fit the rest of the unsused area
gui.rubber(self.controlArea)
self.controlArea.setSizePolicy(
QSizePolicy.Preferred, QSizePolicy.Expanding)
# MAIN AREA
self.scene = QGraphicsScene(self)
self.scene.selectionChanged.connect(self.commit)
self.grid = OWGrid()
self.grid.geometryChanged.connect(self._update_scene_rect)
self.scene.addItem(self.grid)
self.view = QGraphicsView(self.scene)
self.view.setRenderHint(QPainter.Antialiasing, True)
self.view.setVerticalScrollBarPolicy(Qt.ScrollBarAlwaysOn)
self.mainArea.layout().addWidget(self.view)
self.resize(800, 500)
self.clear()
def set_rf(self, model=None):
"""When a different forest is given."""
self.clear()
self.model = model
if model is not None:
if isinstance(model, RandomForestClassifier):
self.forest_type = self.CLASSIFICATION
elif isinstance(model, RandomForestRegressor):
self.forest_type = self.REGRESSION
else:
raise RuntimeError('Invalid type of forest.')
self.forest_adapter = self._get_forest_adapter(self.model)
self.color_palette = self._type_specific('_get_color_palette')()
self._draw_trees()
self.dataset = model.instances
# this bit is important for the regression classifier
if self.dataset is not None and \
self.dataset.domain != model.domain:
self.clf_dataset = Table.from_table(
self.model.domain, self.dataset)
else:
self.clf_dataset = self.dataset
self._update_info_box()
self._type_specific('_update_target_class_combo')()
self._update_depth_slider()
self.selected_tree_index = -1
def clear(self):
"""Clear all relevant data from the widget."""
self.model = None
self.forest_adapter = None
self.ptrees = []
self.grid_items = []
self.grid.clear()
self._clear_info_box()
self._clear_target_class_combo()
self._clear_depth_slider()
# CONTROL AREA CALLBACKS
def max_depth_changed(self):
"""When the max depth slider is changed."""
for tree in self.ptrees:
tree.set_depth_limit(self.depth_limit)
def target_colors_changed(self):
"""When the target class or coloring method is changed."""
for tree in self.ptrees:
tree.target_class_has_changed()
def size_calc_changed(self):
"""When the size calculation of the trees is changed."""
if self.model is not None:
self.forest_adapter = self._get_forest_adapter(self.model)
self.grid.clear()
self._draw_trees()
# Keep the selected item
if self.selected_tree_index != -1:
self.grid_items[self.selected_tree_index].setSelected(True)
self.max_depth_changed()
def zoom_changed(self):
"""When we update the "Zoom" slider."""
for item in self.grid_items:
item.set_max_size(self._calculate_zoom(self.zoom))
width = (self.view.width() - self.view.verticalScrollBar().width())
self.grid.reflow(width)
self.grid.setPreferredWidth(width)
# MODEL CHANGED METHODS
def _update_info_box(self):
self.ui_info.setText(
'Trees: {}'.format(len(self.forest_adapter.get_trees()))
)
def _update_depth_slider(self):
self.depth_limit = self._get_max_depth()
self.ui_depth_slider.parent().setEnabled(True)
self.ui_depth_slider.setMaximum(self.depth_limit)
self.ui_depth_slider.setValue(self.depth_limit)
# MODEL CLEARED METHODS
def _clear_info_box(self):
self.ui_info.setText('No forest on input.')
def _clear_target_class_combo(self):
self.ui_target_class_combo.clear()
self.target_class_index = 0
self.ui_target_class_combo.setCurrentIndex(self.target_class_index)
def _clear_depth_slider(self):
self.ui_depth_slider.parent().setEnabled(False)
self.ui_depth_slider.setMaximum(0)
# HELPFUL METHODS
def _get_max_depth(self):
return max([tree.tree_adapter.max_depth for tree in self.ptrees])
def _get_forest_adapter(self, model):
return SklRandomForestAdapter(model)
def _draw_trees(self):
self.ui_size_calc_combo.setEnabled(False)
self.grid_items, self.ptrees = [], []
with self.progressBar(len(self.forest_adapter.get_trees())) as prg:
for tree in self.forest_adapter.get_trees():
ptree = PythagorasTreeViewer(
None, tree,
node_color_func=self._type_specific('_get_node_color'),
interactive=False, padding=100)
self.grid_items.append(GridItem(
ptree, self.grid, max_size=self._calculate_zoom(self.zoom)
))
self.ptrees.append(ptree)
prg.advance()
self.grid.set_items(self.grid_items)
# This is necessary when adding items for the first time
if self.grid:
width = (self.view.width() -
self.view.verticalScrollBar().width())
self.grid.reflow(width)
self.grid.setPreferredWidth(width)
self.ui_size_calc_combo.setEnabled(True)
@staticmethod
def _calculate_zoom(zoom_level):
"""Calculate the max size for grid items from zoom level setting."""
return zoom_level * 5
def onDeleteWidget(self):
"""When deleting the widget."""
super().onDeleteWidget()
self.clear()
def commit(self):
"""Commit the selected tree to output."""
if len(self.scene.selectedItems()) == 0:
self.send('Tree', None)
# The selected tree index should only reset when model changes
if self.model is None:
self.selected_tree_index = -1
return
selected_item = self.scene.selectedItems()[0]
self.selected_tree_index = self.grid_items.index(selected_item)
obj = self.model.trees[self.selected_tree_index]
obj.instances = self.dataset
obj.meta_target_class_index = self.target_class_index
obj.meta_size_calc_idx = self.size_calc_idx
obj.meta_size_log_scale = self.size_log_scale
obj.meta_depth_limit = self.depth_limit
self.send('Tree', obj)
def send_report(self):
"""Send report."""
self.report_plot()
def _update_scene_rect(self):
self.scene.setSceneRect(self.scene.itemsBoundingRect())
def resizeEvent(self, ev):
width = (self.view.width() - self.view.verticalScrollBar().width())
self.grid.reflow(width)
self.grid.setPreferredWidth(width)
super().resizeEvent(ev)
def _type_specific(self, method):
"""A best effort method getter that somewhat separates logic specific
to classification and regression trees.
This relies on conventional naming of specific methods, e.g.
a method name _get_tooltip would need to be defined like so:
_classification_get_tooltip and _regression_get_tooltip, since they are
both specific.
Parameters
----------
method : str
Method name that we would like to call.
Returns
-------
callable or None
"""
if self.forest_type == self.CLASSIFICATION:
return getattr(self, '_classification' + method)
elif self.forest_type == self.REGRESSION:
return getattr(self, '_regression' + method)
else:
return None
# CLASSIFICATION FOREST SPECIFIC METHODS
def _classification_update_target_class_combo(self):
self._clear_target_class_combo()
self.ui_target_class_combo.addItem('None')
values = [c.title() for c in
self.model.domain.class_vars[0].values]
self.ui_target_class_combo.addItems(values)
def _classification_get_color_palette(self):
return [QColor(*c) for c in self.model.domain.class_var.colors]
def _classification_get_node_color(self, adapter, tree_node):
# this is taken almost directly from the existing classification tree
# viewer
colors = self.color_palette
distribution = adapter.get_distribution(tree_node.label)[0]
total = np.sum(distribution)
if self.target_class_index:
p = distribution[self.target_class_index - 1] / total
color = colors[self.target_class_index - 1].lighter(200 - 100 * p)
else:
modus = np.argmax(distribution)
p = distribution[modus] / (total or 1)
color = colors[int(modus)].lighter(400 - 300 * p)
return color
# REGRESSION FOREST SPECIFIC METHODS
def _regression_update_target_class_combo(self):
self._clear_target_class_combo()
self.ui_target_class_combo.addItems(
list(zip(*self.REGRESSION_COLOR_CALC))[0])
self.ui_target_class_combo.setCurrentIndex(self.target_class_index)
def _regression_get_color_palette(self):
return ContinuousPaletteGenerator(
*self.forest_adapter.domain.class_var.colors)
def _regression_get_node_color(self, adapter, tree_node):
return self.REGRESSION_COLOR_CALC[self.target_class_index][1](
adapter, tree_node
)
def _color_class_mean(self, adapter, tree_node):
# calculate node colors relative to the mean of the node samples
min_mean = np.min(self.clf_dataset.Y)
max_mean = np.max(self.clf_dataset.Y)
instances = adapter.get_instances_in_nodes(self.clf_dataset,
tree_node.label)
mean = np.mean(instances.Y)
return self.color_palette[(mean - min_mean) / (max_mean - min_mean)]
def _color_stddev(self, adapter, tree_node):
# calculate node colors relative to the standard deviation in the node
# samples
min_mean, max_mean = 0, np.std(self.clf_dataset.Y)
instances = adapter.get_instances_in_nodes(self.clf_dataset,
tree_node.label)
std = np.std(instances.Y)
return self.color_palette[(std - min_mean) / (max_mean - min_mean)]
class OWQualityControl(widget.OWWidget):
name = "Quality Control"
description = "Experiment quality control"
icon = "../widgets/icons/QualityControl.svg"
priority = 5000
inputs = [("Experiment Data", Orange.data.Table, "set_data")]
outputs = []
DISTANCE_FUNCTIONS = [("Distance from Pearson correlation",
dist_pcorr),
("Euclidean distance",
dist_eucl),
("Distance from Spearman correlation",
dist_spearman)]
settingsHandler = SetContextHandler()
split_by_labels = settings.ContextSetting({})
sort_by_labels = settings.ContextSetting({})
selected_distance_index = settings.Setting(0)
def __init__(self, parent=None):
super().__init__(parent)
## Attributes
self.data = None
self.distances = None
self.groups = None
self.unique_pos = None
self.base_group_index = 0
## GUI
box = gui.widgetBox(self.controlArea, "Info")
self.info_box = gui.widgetLabel(box, "\n")
## Separate By box
box = gui.widgetBox(self.controlArea, "Separate By")
self.split_by_model = itemmodels.PyListModel(parent=self)
self.split_by_view = QListView()
self.split_by_view.setSelectionMode(QListView.ExtendedSelection)
self.split_by_view.setModel(self.split_by_model)
box.layout().addWidget(self.split_by_view)
self.split_by_view.selectionModel().selectionChanged.connect(
self.on_split_key_changed)
## Sort By box
box = gui.widgetBox(self.controlArea, "Sort By")
self.sort_by_model = itemmodels.PyListModel(parent=self)
self.sort_by_view = QListView()
self.sort_by_view.setSelectionMode(QListView.ExtendedSelection)
self.sort_by_view.setModel(self.sort_by_model)
box.layout().addWidget(self.sort_by_view)
self.sort_by_view.selectionModel().selectionChanged.connect(
self.on_sort_key_changed)
## Distance box
box = gui.widgetBox(self.controlArea, "Distance Measure")
gui.comboBox(box, self, "selected_distance_index",
items=[name for name, _ in self.DISTANCE_FUNCTIONS],
callback=self.on_distance_measure_changed)
self.scene = QGraphicsScene()
self.scene_view = QGraphicsView(self.scene)
self.scene_view.setRenderHints(QPainter.Antialiasing)
self.scene_view.setAlignment(Qt.AlignLeft | Qt.AlignVCenter)
self.mainArea.layout().addWidget(self.scene_view)
self.scene_view.installEventFilter(self)
self._disable_updates = False
self._cached_distances = {}
self._base_index_hints = {}
self.main_widget = None
self.resize(800, 600)
def clear(self):
"""Clear the widget state."""
self.data = None
self.distances = None
self.groups = None
self.unique_pos = None
with disable_updates(self):
self.split_by_model[:] = []
self.sort_by_model[:] = []
self.main_widget = None
self.scene.clear()
self.info_box.setText("\n")
self._cached_distances = {}
def set_data(self, data=None):
"""Set input experiment data."""
self.closeContext()
self.clear()
self.error(0)
self.warning(0)
if data is not None:
keys = self.get_suitable_keys(data)
if not keys:
self.error(0, "Data has no suitable feature labels.")
data = None
self.data = data
if data is not None:
self.on_new_data()
def update_label_candidates(self):
"""Update the label candidates selection GUI
(Group/Sort By views).
"""
keys = self.get_suitable_keys(self.data)
with disable_updates(self):
self.split_by_model[:] = keys
self.sort_by_model[:] = keys
def get_suitable_keys(self, data):
""" Return suitable attr label keys from the data where
the key has at least two unique values in the data.
"""
attrs = [attr.attributes.items() for attr in data.domain.attributes]
attrs = reduce(operator.iadd, attrs, [])
# in case someone put non string values in attributes dict
attrs = [(str(key), str(value)) for key, value in attrs]
attrs = set(attrs)
values = defaultdict(set)
for key, value in attrs:
values[key].add(value)
keys = [key for key in values if len(values[key]) > 1]
return keys
def selected_split_by_labels(self):
"""Return the current selected split labels.
"""
sel_m = self.split_by_view.selectionModel()
indices = [r.row() for r in sel_m.selectedRows()]
return [self.sort_by_model[i] for i in indices]
def selected_sort_by_labels(self):
"""Return the current selected sort labels
"""
sel_m = self.sort_by_view.selectionModel()
indices = [r.row() for r in sel_m.selectedRows()]
return [self.sort_by_model[i] for i in indices]
def selected_distance(self):
"""Return the selected distance function.
"""
return self.DISTANCE_FUNCTIONS[self.selected_distance_index][1]
def selected_base_group_index(self):
"""Return the selected base group index
"""
return self.base_group_index
def selected_base_indices(self, base_group_index=None):
indices = []
for g, ind in self.groups:
if base_group_index is None:
label = group_label(self.selected_split_by_labels(), g)
ind = [i for i in ind if i is not None]
i = self._base_index_hints.get(label, ind[0] if ind else None)
else:
i = ind[base_group_index]
indices.append(i)
return indices
def on_new_data(self):
"""We have new data and need to recompute all.
"""
self.closeContext()
self.update_label_candidates()
self.info_box.setText(
"%s genes \n%s experiments" %
(len(self.data), len(self.data.domain.attributes))
)
self.base_group_index = 0
keys = self.get_suitable_keys(self.data)
self.openContext(keys)
## Restore saved context settings (split/sort selection)
split_by_labels = self.split_by_labels
sort_by_labels = self.sort_by_labels
def select(model, selection_model, selected_items):
"""Select items in a Qt item model view
"""
all_items = list(model)
try:
indices = [all_items.index(item) for item in selected_items]
except:
indices = []
for ind in indices:
selection_model.select(model.index(ind),
QItemSelectionModel.Select)
with disable_updates(self):
select(self.split_by_view.model(),
self.split_by_view.selectionModel(),
split_by_labels)
select(self.sort_by_view.model(),
self.sort_by_view.selectionModel(),
sort_by_labels)
with widget_disable(self):
self.split_and_update()
def on_split_key_changed(self, *args):
"""Split key has changed
"""
with widget_disable(self):
if not self._disable_updates:
self.base_group_index = 0
self.split_by_labels = self.selected_split_by_labels()
self.split_and_update()
def on_sort_key_changed(self, *args):
"""Sort key has changed
"""
with widget_disable(self):
if not self._disable_updates:
self.base_group_index = 0
self.sort_by_labels = self.selected_sort_by_labels()
self.split_and_update()
def on_distance_measure_changed(self):
"""Distance measure has changed
"""
if self.data is not None:
with widget_disable(self):
self.update_distances()
self.replot_experiments()
def on_view_resize(self, size):
"""The view with the quality plot has changed
"""
if self.main_widget:
current = self.main_widget.size()
self.main_widget.resize(size.width() - 6,
current.height())
self.scene.setSceneRect(self.scene.itemsBoundingRect())
def on_rug_item_clicked(self, item):
"""An ``item`` in the quality plot has been clicked.
"""
update = False
sort_by_labels = self.selected_sort_by_labels()
if sort_by_labels and item.in_group:
## The item is part of the group
if item.group_index != self.base_group_index:
self.base_group_index = item.group_index
update = True
else:
if sort_by_labels:
# If the user clicked on an background item it
# invalidates the sorted labels selection
with disable_updates(self):
self.sort_by_view.selectionModel().clear()
update = True
index = item.index
group = item.group
label = group_label(self.selected_split_by_labels(), group)
if self._base_index_hints.get(label, 0) != index:
self._base_index_hints[label] = index
update = True
if update:
with widget_disable(self):
self.split_and_update()
def eventFilter(self, obj, event):
if obj is self.scene_view and event.type() == QEvent.Resize:
self.on_view_resize(event.size())
return super().eventFilter(obj, event)
def split_and_update(self):
"""
Split the data based on the selected sort/split labels
and update the quality plot.
"""
split_labels = self.selected_split_by_labels()
sort_labels = self.selected_sort_by_labels()
self.warning(0)
if not split_labels:
self.warning(0, "No separate by label selected.")
self.groups, self.unique_pos = \
exp.separate_by(self.data, split_labels,
consider=sort_labels,
add_empty=True)
self.groups = sorted(self.groups.items(),
key=lambda t: list(map(float_if_posible, t[0])))
self.unique_pos = sorted(self.unique_pos.items(),
key=lambda t: list(map(float_if_posible, t[0])))
if self.groups:
if sort_labels:
group_base = self.selected_base_group_index()
base_indices = self.selected_base_indices(group_base)
else:
base_indices = self.selected_base_indices()
self.update_distances(base_indices)
self.replot_experiments()
def get_cached_distances(self, measure):
if measure not in self._cached_distances:
attrs = self.data.domain.attributes
mat = numpy.zeros((len(attrs), len(attrs)))
self._cached_distances[measure] = \
(mat, set(zip(range(len(attrs)), range(len(attrs)))))
return self._cached_distances[measure]
def get_cached_distance(self, measure, i, j):
matrix, computed = self.get_cached_distances(measure)
key = (i, j) if i < j else (j, i)
if key in computed:
return matrix[i, j]
else:
return None
def get_distance(self, measure, i, j):
d = self.get_cached_distance(measure, i, j)
if d is None:
vec_i = take_columns(self.data, [i])
vec_j = take_columns(self.data, [j])
d = measure(vec_i, vec_j)
mat, computed = self.get_cached_distances(measure)
mat[i, j] = d
key = key = (i, j) if i < j else (j, i)
computed.add(key)
return d
def store_distance(self, measure, i, j, dist):
matrix, computed = self.get_cached_distances(measure)
key = (i, j) if i < j else (j, i)
matrix[j, i] = matrix[i, j] = dist
computed.add(key)
def update_distances(self, base_indices=()):
"""Recompute the experiment distances.
"""
distance = self.selected_distance()
if base_indices == ():
base_group_index = self.selected_base_group_index()
base_indices = [ind[base_group_index] \
for _, ind in self.groups]
assert(len(base_indices) == len(self.groups))
base_distances = []
attributes = self.data.domain.attributes
pb = gui.ProgressBar(self, len(self.groups) * len(attributes))
for (group, indices), base_index in zip(self.groups, base_indices):
# Base column of the group
if base_index is not None:
base_vec = take_columns(self.data, [base_index])
distances = []
# Compute the distances between base column
# and all the rest data columns.
for i in range(len(attributes)):
if i == base_index:
distances.append(0.0)
elif self.get_cached_distance(distance, i, base_index) is not None:
distances.append(self.get_cached_distance(distance, i, base_index))
else:
vec_i = take_columns(self.data, [i])
dist = distance(base_vec, vec_i)
self.store_distance(distance, i, base_index, dist)
distances.append(dist)
pb.advance()
base_distances.append(distances)
else:
base_distances.append(None)
pb.finish()
self.distances = base_distances
def replot_experiments(self):
"""Replot the whole quality plot.
"""
self.scene.clear()
labels = []
max_dist = numpy.nanmax(list(filter(None, self.distances)))
rug_widgets = []
group_pen = QPen(Qt.black)
group_pen.setWidth(2)
group_pen.setCapStyle(Qt.RoundCap)
background_pen = QPen(QColor(0, 0, 250, 150))
background_pen.setWidth(1)
background_pen.setCapStyle(Qt.RoundCap)
main_widget = QGraphicsWidget()
layout = QGraphicsGridLayout()
attributes = self.data.domain.attributes
if self.data is not None:
for (group, indices), dist_vec in zip(self.groups, self.distances):
indices_set = set(indices)
rug_items = []
if dist_vec is not None:
for i, attr in enumerate(attributes):
# Is this a within group distance or background
in_group = i in indices_set
if in_group:
rug_item = ClickableRugItem(dist_vec[i] / max_dist,
1.0, self.on_rug_item_clicked)
rug_item.setPen(group_pen)
tooltip = experiment_description(attr)
rug_item.setToolTip(tooltip)
rug_item.group_index = indices.index(i)
rug_item.setZValue(rug_item.zValue() + 1)
else:
rug_item = ClickableRugItem(dist_vec[i] / max_dist,
0.85, self.on_rug_item_clicked)
rug_item.setPen(background_pen)
tooltip = experiment_description(attr)
rug_item.setToolTip(tooltip)
rug_item.group = group
rug_item.index = i
rug_item.in_group = in_group
rug_items.append(rug_item)
rug_widget = RugGraphicsWidget(parent=main_widget)
rug_widget.set_rug(rug_items)
rug_widgets.append(rug_widget)
label = group_label(self.selected_split_by_labels(), group)
label_item = QGraphicsSimpleTextItem(label, main_widget)
label_item = GraphicsSimpleTextLayoutItem(label_item, parent=layout)
label_item.setSizePolicy(QSizePolicy.Fixed, QSizePolicy.Fixed)
labels.append(label_item)
for i, (label, rug_w) in enumerate(zip(labels, rug_widgets)):
layout.addItem(label, i, 0, Qt.AlignVCenter)
layout.addItem(rug_w, i, 1)
layout.setRowMaximumHeight(i, 30)
main_widget.setLayout(layout)
self.scene.addItem(main_widget)
self.main_widget = main_widget
self.rug_widgets = rug_widgets
self.labels = labels
self.on_view_resize(self.scene_view.size())
class OWSOM(OWWidget):
name = "Self-Organizing Map"
description = "Computation of self-organizing map."
icon = "icons/SOM.svg"
keywords = ["SOM"]
class Inputs:
data = Input("Data", Table)
class Outputs:
selected_data = Output("Selected Data", Table, default=True)
annotated_data = Output(ANNOTATED_DATA_SIGNAL_NAME, Table)
settingsHandler = DomainContextHandler()
auto_dimension = Setting(True)
size_x = Setting(10)
size_y = Setting(10)
hexagonal = Setting(1)
initialization = Setting(0)
attr_color = ContextSetting(None)
size_by_instances = Setting(True)
pie_charts = Setting(False)
selection = Setting(None, schema_only=True)
graph_name = "view"
_grid_pen = QPen(QBrush(QColor(224, 224, 224)), 2)
_grid_pen.setCosmetic(True)
OptControls = namedtuple(
"OptControls",
("shape", "auto_dim", "spin_x", "spin_y", "initialization", "start"))
class Warning(OWWidget.Warning):
ignoring_disc_variables = Msg("SOM ignores discrete variables.")
missing_colors = \
Msg("Some data instances have undefined value of '{}'.")
missing_values = \
Msg("{} data instance{} with undefined value(s) {} not shown.")
single_attribute = Msg("Data contains a single numeric column.")
class Error(OWWidget.Error):
no_numeric_variables = Msg("Data contains no numeric columns.")
no_defined_rows = Msg("All rows contain at least one undefined value.")
def __init__(self):
super().__init__()
self.__pending_selection = self.selection
self._optimizer = None
self._optimizer_thread = None
self.stop_optimization = False
self.data = self.cont_x = None
self.cells = self.member_data = None
self.selection = None
self.colors = self.thresholds = self.bin_labels = None
self._set_input_summary(None)
self._set_output_summary(None)
box = gui.vBox(self.controlArea, box="SOM")
shape = gui.comboBox(box,
self,
"",
items=("Hexagonal grid", "Square grid"))
shape.setCurrentIndex(1 - self.hexagonal)
box2 = gui.indentedBox(box, 10)
auto_dim = gui.checkBox(box2,
self,
"auto_dimension",
"Set dimensions automatically",
callback=self.on_auto_dimension_changed)
self.manual_box = box3 = gui.hBox(box2)
spinargs = dict(value="",
widget=box3,
master=self,
minv=5,
maxv=100,
step=5,
alignment=Qt.AlignRight)
spin_x = gui.spin(**spinargs)
spin_x.setValue(self.size_x)
gui.widgetLabel(box3, "×")
spin_y = gui.spin(**spinargs)
spin_y.setValue(self.size_y)
gui.rubber(box3)
self.manual_box.setEnabled(not self.auto_dimension)
initialization = gui.comboBox(box,
self,
"initialization",
items=("Initialize with PCA",
"Random initialization",
"Replicable random"))
start = gui.button(box,
self,
"Restart",
callback=self.restart_som_pressed,
sizePolicy=(QSizePolicy.MinimumExpanding,
QSizePolicy.Fixed))
self.opt_controls = self.OptControls(shape, auto_dim, spin_x, spin_y,
initialization, start)
box = gui.vBox(self.controlArea, "Color")
gui.comboBox(box,
self,
"attr_color",
searchable=True,
callback=self.on_attr_color_change,
model=DomainModel(placeholder="(Same color)",
valid_types=DomainModel.PRIMITIVE))
gui.checkBox(box,
self,
"pie_charts",
label="Show pie charts",
callback=self.on_pie_chart_change)
gui.checkBox(box,
self,
"size_by_instances",
label="Size by number of instances",
callback=self.on_attr_size_change)
gui.rubber(self.controlArea)
self.scene = QGraphicsScene(self)
self.view = SomView(self.scene)
self.view.setMinimumWidth(400)
self.view.setMinimumHeight(400)
self.view.setHorizontalScrollBarPolicy(Qt.ScrollBarAlwaysOff)
self.view.setVerticalScrollBarPolicy(Qt.ScrollBarAlwaysOff)
self.view.setRenderHint(QPainter.Antialiasing)
self.view.selection_changed.connect(self.on_selection_change)
self.view.selection_moved.connect(self.on_selection_move)
self.view.selection_mark_changed.connect(self.on_selection_mark_change)
self.mainArea.layout().addWidget(self.view)
self.elements = None
self.grid = None
self.grid_cells = None
self.legend = None
@Inputs.data
def set_data(self, data):
def prepare_data():
if len(cont_attrs) < len(attrs):
self.Warning.ignoring_disc_variables()
if len(cont_attrs) == 1:
self.Warning.single_attribute()
x = Table.from_table(Domain(cont_attrs), data).X
if sp.issparse(x):
self.data = data
self.cont_x = x.tocsr()
else:
mask = np.all(np.isfinite(x), axis=1)
if not np.any(mask):
self.Error.no_defined_rows()
else:
if np.all(mask):
self.data = data
self.cont_x = x.copy()
else:
self.data = data[mask]
self.cont_x = x[mask]
self.cont_x -= np.min(self.cont_x, axis=0)[None, :]
sums = np.sum(self.cont_x, axis=0)[None, :]
sums[sums == 0] = 1
self.cont_x /= sums
def set_warnings():
missing = len(data) - len(self.data)
if missing == 1:
self.Warning.missing_values(1, "", "is")
elif missing > 1:
self.Warning.missing_values(missing, "s", "are")
self.stop_optimization_and_wait()
self.closeContext()
self.clear()
self.Error.clear()
self.Warning.clear()
if data is not None:
attrs = data.domain.attributes
cont_attrs = [var for var in attrs if var.is_continuous]
if not cont_attrs:
self.Error.no_numeric_variables()
else:
prepare_data()
if self.data is not None:
self.controls.attr_color.model().set_domain(data.domain)
self.attr_color = data.domain.class_var
set_warnings()
self.openContext(self.data)
self.set_color_bins()
self.create_legend()
self.recompute_dimensions()
self._set_input_summary(data)
self.start_som()
def _set_input_summary(self, data):
summary = len(data) if data else self.info.NoInput
details = format_summary_details(data) if data else ""
self.info.set_input_summary(summary, details)
def _set_output_summary(self, output):
summary = len(output) if output else self.info.NoOutput
details = format_summary_details(output) if output else ""
self.info.set_output_summary(summary, details)
def clear(self):
self.data = self.cont_x = None
self.cells = self.member_data = None
self.attr_color = None
self.colors = self.thresholds = self.bin_labels = None
if self.elements is not None:
self.scene.removeItem(self.elements)
self.elements = None
self.clear_selection()
self.controls.attr_color.model().set_domain(None)
self.Warning.clear()
self.Error.clear()
def recompute_dimensions(self):
if not self.auto_dimension or self.cont_x is None:
return
dim = max(5, int(np.ceil(np.sqrt(5 * np.sqrt(self.cont_x.shape[0])))))
self.opt_controls.spin_x.setValue(dim)
self.opt_controls.spin_y.setValue(dim)
def on_auto_dimension_changed(self):
self.manual_box.setEnabled(not self.auto_dimension)
if self.auto_dimension:
self.recompute_dimensions()
else:
spin_x = self.opt_controls.spin_x
spin_y = self.opt_controls.spin_y
dimx = int(5 * np.round(spin_x.value() / 5))
dimy = int(5 * np.round(spin_y.value() / 5))
spin_x.setValue(dimx)
spin_y.setValue(dimy)
def on_attr_color_change(self):
self.controls.pie_charts.setEnabled(self.attr_color is not None)
self.set_color_bins()
self.create_legend()
self.rescale()
self._redraw()
def on_attr_size_change(self):
self._redraw()
def on_pie_chart_change(self):
self._redraw()
def clear_selection(self):
self.selection = None
self.redraw_selection()
def on_selection_change(self, selection, action=SomView.SelectionSet):
if self.data is None: # clicks on empty canvas
return
if self.selection is None:
self.selection = np.zeros(self.grid_cells.T.shape, dtype=np.int16)
if action == SomView.SelectionSet:
self.selection[:] = 0
self.selection[selection] = 1
elif action == SomView.SelectionAddToGroup:
self.selection[selection] = max(1, np.max(self.selection))
elif action == SomView.SelectionNewGroup:
self.selection[selection] = 1 + np.max(self.selection)
elif action & SomView.SelectionRemove:
self.selection[selection] = 0
self.redraw_selection()
self.update_output()
def on_selection_move(self, event: QKeyEvent):
if self.selection is None or not np.any(self.selection):
if event.key() in (Qt.Key_Right, Qt.Key_Down):
x = y = 0
else:
x = self.size_x - 1
y = self.size_y - 1
else:
x, y = np.nonzero(self.selection)
if len(x) > 1:
return
if event.key() == Qt.Key_Up and y > 0:
y -= 1
if event.key() == Qt.Key_Down and y < self.size_y - 1:
y += 1
if event.key() == Qt.Key_Left and x:
x -= 1
if event.key() == Qt.Key_Right and x < self.size_x - 1:
x += 1
x -= self.hexagonal and x == self.size_x - 1 and y % 2
if self.selection is not None and self.selection[x, y]:
return
selection = np.zeros(self.grid_cells.shape, dtype=bool)
selection[x, y] = True
self.on_selection_change(selection)
def on_selection_mark_change(self, marks):
self.redraw_selection(marks=marks)
def redraw_selection(self, marks=None):
if self.grid_cells is None:
return
sel_pen = QPen(QBrush(QColor(128, 128, 128)), 2)
sel_pen.setCosmetic(True)
mark_pen = QPen(QBrush(QColor(128, 128, 128)), 4)
mark_pen.setCosmetic(True)
pens = [self._grid_pen, sel_pen]
mark_brush = QBrush(QColor(224, 255, 255))
sels = self.selection is not None and np.max(self.selection)
palette = LimitedDiscretePalette(number_of_colors=sels + 1)
brushes = [QBrush(Qt.NoBrush)] + \
[QBrush(palette[i].lighter(165)) for i in range(sels)]
for y in range(self.size_y):
for x in range(self.size_x - (y % 2) * self.hexagonal):
cell = self.grid_cells[y, x]
marked = marks is not None and marks[x, y]
sel_group = self.selection is not None and self.selection[x, y]
if marked:
cell.setBrush(mark_brush)
cell.setPen(mark_pen)
else:
cell.setBrush(brushes[sel_group])
cell.setPen(pens[bool(sel_group)])
cell.setZValue(marked or sel_group)
def restart_som_pressed(self):
if self._optimizer_thread is not None:
self.stop_optimization = True
self._optimizer.stop_optimization = True
else:
self.start_som()
def start_som(self):
self.read_controls()
self.update_layout()
self.clear_selection()
if self.cont_x is not None:
self.enable_controls(False)
self._recompute_som()
else:
self.update_output()
def read_controls(self):
c = self.opt_controls
self.hexagonal = c.shape.currentIndex() == 0
self.size_x = c.spin_x.value()
self.size_y = c.spin_y.value()
def enable_controls(self, enable):
c = self.opt_controls
c.shape.setEnabled(enable)
c.auto_dim.setEnabled(enable)
c.start.setText("Start" if enable else "Stop")
def update_layout(self):
self.set_legend_pos()
if self.elements: # Prevent having redrawn grid but with old elements
self.scene.removeItem(self.elements)
self.elements = None
self.redraw_grid()
self.rescale()
def _redraw(self):
self.Warning.missing_colors.clear()
if self.elements:
self.scene.removeItem(self.elements)
self.elements = None
self.view.set_dimensions(self.size_x, self.size_y, self.hexagonal)
if self.cells is None:
return
sizes = self.cells[:, :, 1] - self.cells[:, :, 0]
sizes = sizes.astype(float)
if not self.size_by_instances:
sizes[sizes != 0] = 0.8
else:
sizes *= 0.8 / np.max(sizes)
self.elements = QGraphicsItemGroup()
self.scene.addItem(self.elements)
if self.attr_color is None:
self._draw_same_color(sizes)
elif self.pie_charts:
self._draw_pie_charts(sizes)
else:
self._draw_colored_circles(sizes)
@property
def _grid_factors(self):
return (0.5, sqrt3_2) if self.hexagonal else (0, 1)
def _draw_same_color(self, sizes):
fx, fy = self._grid_factors
color = QColor(64, 64, 64)
for y in range(self.size_y):
for x in range(self.size_x - self.hexagonal * (y % 2)):
r = sizes[x, y]
n = len(self.get_member_indices(x, y))
if not r:
continue
ellipse = ColoredCircle(r / 2, color, 0)
ellipse.setPos(x + (y % 2) * fx, y * fy)
ellipse.setToolTip(f"{n} instances")
self.elements.addToGroup(ellipse)
def _get_color_column(self):
color_column = \
self.data.get_column_view(self.attr_color)[0].astype(float,
copy=False)
if self.attr_color.is_discrete:
with np.errstate(invalid="ignore"):
int_col = color_column.astype(int)
int_col[np.isnan(color_column)] = len(self.colors)
else:
int_col = np.zeros(len(color_column), dtype=int)
# The following line is unnecessary because rows with missing
# numeric data are excluded. Uncomment it if you change SOM to
# tolerate missing values.
# int_col[np.isnan(color_column)] = len(self.colors)
for i, thresh in enumerate(self.thresholds, start=1):
int_col[color_column >= thresh] = i
return int_col
def _tooltip(self, colors, distribution):
if self.attr_color.is_discrete:
values = self.attr_color.values
else:
values = self._bin_names()
tot = np.sum(distribution)
nbhp = "\N{NON-BREAKING HYPHEN}"
return '<table style="white-space: nowrap">' + "".join(f"""
<tr>
<td>
<font color={color.name()}>■</font>
<b>{escape(val).replace("-", nbhp)}</b>:
</td>
<td>
{n} ({n / tot * 100:.1f} %)
</td>
</tr>
""" for color, val, n in zip(colors, values, distribution) if n) \
+ "</table>"
def _draw_pie_charts(self, sizes):
fx, fy = self._grid_factors
color_column = self._get_color_column()
colors = self.colors.qcolors_w_nan
for y in range(self.size_y):
for x in range(self.size_x - self.hexagonal * (y % 2)):
r = sizes[x, y]
if not r:
self.grid_cells[y, x].setToolTip("")
continue
members = self.get_member_indices(x, y)
color_dist = np.bincount(color_column[members],
minlength=len(colors))
rel_color_dist = color_dist.astype(float) / len(members)
pie = PieChart(rel_color_dist, r / 2, colors)
pie.setToolTip(self._tooltip(colors, color_dist))
self.elements.addToGroup(pie)
pie.setPos(x + (y % 2) * fx, y * fy)
def _draw_colored_circles(self, sizes):
fx, fy = self._grid_factors
color_column = self._get_color_column()
qcolors = self.colors.qcolors_w_nan
for y in range(self.size_y):
for x in range(self.size_x - self.hexagonal * (y % 2)):
r = sizes[x, y]
if not r:
continue
members = self.get_member_indices(x, y)
color_dist = color_column[members]
color_dist = color_dist[color_dist < len(self.colors)]
if len(color_dist) != len(members):
self.Warning.missing_colors(self.attr_color.name)
bc = np.bincount(color_dist, minlength=len(self.colors))
color = qcolors[np.argmax(bc)]
ellipse = ColoredCircle(r / 2, color,
np.max(bc) / len(members))
ellipse.setPos(x + (y % 2) * fx, y * fy)
ellipse.setToolTip(self._tooltip(qcolors, bc))
self.elements.addToGroup(ellipse)
def redraw_grid(self):
if self.grid is not None:
self.scene.removeItem(self.grid)
self.grid = QGraphicsItemGroup()
self.grid.setZValue(-200)
self.grid_cells = np.full((self.size_y, self.size_x), None)
for y in range(self.size_y):
for x in range(self.size_x - (y % 2) * self.hexagonal):
if self.hexagonal:
cell = QGraphicsPathItem(_hexagon_path)
cell.setPos(x + (y % 2) / 2, y * sqrt3_2)
else:
cell = QGraphicsRectItem(x - 0.5, y - 0.5, 1, 1)
self.grid_cells[y, x] = cell
cell.setPen(self._grid_pen)
self.grid.addToGroup(cell)
self.scene.addItem(self.grid)
def get_member_indices(self, x, y):
i, j = self.cells[x, y]
return self.member_data[i:j]
def _recompute_som(self):
if self.cont_x is None:
return
som = SOM(self.size_x,
self.size_y,
hexagonal=self.hexagonal,
pca_init=self.initialization == 0,
random_seed=0 if self.initialization == 2 else None)
class Optimizer(QObject):
update = Signal(float, np.ndarray, np.ndarray)
done = Signal(SOM)
stopped = Signal()
stop_optimization = False
def __init__(self, data, som):
super().__init__()
self.som = som
self.data = data
def callback(self, progress):
self.update.emit(progress, self.som.weights.copy(),
self.som.ssum_weights.copy())
return not self.stop_optimization
def run(self):
try:
self.som.fit(self.data,
N_ITERATIONS,
callback=self.callback)
# Report an exception, but still remove the thread
finally:
self.done.emit(self.som)
self.stopped.emit()
def thread_finished():
self._optimizer = None
self._optimizer_thread = None
self.progressBarInit()
self._optimizer = Optimizer(self.cont_x, som)
self._optimizer_thread = QThread()
self._optimizer_thread.setStackSize(5 * 2**20)
self._optimizer.update.connect(self.__update)
self._optimizer.done.connect(self.__done)
self._optimizer.stopped.connect(self._optimizer_thread.quit)
self._optimizer.moveToThread(self._optimizer_thread)
self._optimizer_thread.started.connect(self._optimizer.run)
self._optimizer_thread.finished.connect(thread_finished)
self.stop_optimization = False
self._optimizer_thread.start()
@Slot(float, object, object)
def __update(self, _progress, weights, ssum_weights):
self.progressBarSet(_progress)
self._assign_instances(weights, ssum_weights)
self._redraw()
@Slot(object)
def __done(self, som):
self.enable_controls(True)
self.progressBarFinished()
self._assign_instances(som.weights, som.ssum_weights)
self._redraw()
# This is the first time we know what was selected (assuming that
# initialization is not set to random)
if self.__pending_selection is not None:
self.on_selection_change(self.__pending_selection)
self.__pending_selection = None
self.update_output()
def stop_optimization_and_wait(self):
if self._optimizer_thread is not None:
self.stop_optimization = True
self._optimizer.stop_optimization = True
self._optimizer_thread.quit()
self._optimizer_thread.wait()
self._optimizer_thread = None
def onDeleteWidget(self):
self.stop_optimization_and_wait()
self.clear()
super().onDeleteWidget()
def _assign_instances(self, weights, ssum_weights):
if self.cont_x is None:
return # the widget is shutting down while signals still processed
assignments = SOM.winner_from_weights(self.cont_x, weights,
ssum_weights, self.hexagonal)
members = defaultdict(list)
for i, (x, y) in enumerate(assignments):
members[(x, y)].append(i)
members.pop(None, None)
self.cells = np.empty((self.size_x, self.size_y, 2), dtype=int)
self.member_data = np.empty(self.cont_x.shape[0], dtype=int)
index = 0
for x in range(self.size_x):
for y in range(self.size_y):
nmembers = len(members[(x, y)])
self.member_data[index:index + nmembers] = members[(x, y)]
self.cells[x, y] = [index, index + nmembers]
index += nmembers
def resizeEvent(self, event):
super().resizeEvent(event)
self.create_legend() # re-wrap lines if necessary
self.rescale()
def rescale(self):
if self.legend:
leg_height = self.legend.boundingRect().height()
leg_extra = 1.5
else:
leg_height = 0
leg_extra = 1
vw, vh = self.view.width(), self.view.height() - leg_height
scale = min(vw / (self.size_x + 1),
vh / ((self.size_y + leg_extra) * self._grid_factors[1]))
self.view.setTransform(QTransform.fromScale(scale, scale))
if self.hexagonal:
self.view.setSceneRect(0, -1, self.size_x - 1,
(self.size_y + leg_extra) * sqrt3_2 +
leg_height / scale)
else:
self.view.setSceneRect(-0.25, -0.25, self.size_x - 0.5,
self.size_y - 0.5 + leg_height / scale)
def update_output(self):
if self.data is None:
self.Outputs.selected_data.send(None)
self.Outputs.annotated_data.send(None)
self._set_output_summary(None)
return
indices = np.zeros(len(self.data), dtype=int)
if self.selection is not None and np.any(self.selection):
for y in range(self.size_y):
for x in range(self.size_x):
rows = self.get_member_indices(x, y)
indices[rows] = self.selection[x, y]
if np.any(indices):
sel_data = create_groups_table(self.data, indices, False, "Group")
self.Outputs.selected_data.send(sel_data)
self._set_output_summary(sel_data)
else:
self.Outputs.selected_data.send(None)
self._set_output_summary(None)
if np.max(indices) > 1:
annotated = create_groups_table(self.data, indices)
else:
annotated = create_annotated_table(self.data,
np.flatnonzero(indices))
self.Outputs.annotated_data.send(annotated)
def set_color_bins(self):
if self.attr_color is None:
self.thresholds = self.bin_labels = self.colors = None
elif self.attr_color.is_discrete:
self.thresholds = self.bin_labels = None
self.colors = self.attr_color.palette
else:
col = self.data.get_column_view(self.attr_color)[0].astype(float)
if self.attr_color.is_time:
binning = time_binnings(col, min_bins=4)[-1]
else:
binning = decimal_binnings(col, min_bins=4)[-1]
self.thresholds = binning.thresholds[1:-1]
self.bin_labels = (binning.labels[1:-1],
binning.short_labels[1:-1])
palette = BinnedContinuousPalette.from_palette(
self.attr_color.palette, binning.thresholds)
self.colors = palette
def create_legend(self):
if self.legend is not None:
self.scene.removeItem(self.legend)
self.legend = None
if self.attr_color is None:
return
if self.attr_color.is_discrete:
names = self.attr_color.values
else:
names = self._bin_names()
items = []
size = 8
for name, color in zip(names, self.colors.qcolors):
item = QGraphicsItemGroup()
item.addToGroup(
CanvasRectangle(None, -size / 2, -size / 2, size, size,
Qt.gray, color))
item.addToGroup(CanvasText(None, name, size, 0, Qt.AlignVCenter))
items.append(item)
self.legend = wrap_legend_items(items,
hspacing=20,
vspacing=16 + size,
max_width=self.view.width() - 25)
self.legend.setFlags(self.legend.ItemIgnoresTransformations)
self.legend.setTransform(
QTransform.fromTranslate(-self.legend.boundingRect().width() / 2,
0))
self.scene.addItem(self.legend)
self.set_legend_pos()
def _bin_names(self):
labels, short_labels = self.bin_labels
return \
[f"< {labels[0]}"] \
+ [f"{x} - {y}" for x, y in zip(labels, short_labels[1:])] \
+ [f"≥ {labels[-1]}"]
def set_legend_pos(self):
if self.legend is None:
return
self.legend.setPos(self.size_x / 2,
(self.size_y + 0.2 + 0.3 * self.hexagonal) *
self._grid_factors[1])
def send_report(self):
self.report_plot()
if self.attr_color:
self.report_caption(
f"Self-organizing map colored by '{self.attr_color.name}'")
class OWSieveDiagram(OWWidget):
name = "Sieve Diagram"
description = "Visualize the observed and expected frequencies " \
"for a combination of values."
icon = "icons/SieveDiagram.svg"
priority = 200
class Inputs:
data = Input("Data", Table, default=True)
features = Input("Features", AttributeList)
class Outputs:
selected_data = Output("Selected Data", Table, default=True)
annotated_data = Output(ANNOTATED_DATA_SIGNAL_NAME, Table)
graph_name = "canvas"
want_control_area = False
settings_version = 1
settingsHandler = DomainContextHandler()
attr_x = ContextSetting(None)
attr_y = ContextSetting(None)
selection = ContextSetting(set())
def __init__(self):
# pylint: disable=missing-docstring
super().__init__()
self.data = self.discrete_data = None
self.attrs = []
self.input_features = None
self.areas = []
self.selection = set()
self.attr_box = gui.hBox(self.mainArea)
self.domain_model = DomainModel(valid_types=DomainModel.PRIMITIVE)
combo_args = dict(
widget=self.attr_box, master=self, contentsLength=12,
callback=self.update_attr, sendSelectedValue=True, valueType=str,
model=self.domain_model)
fixed_size = (QSizePolicy.Fixed, QSizePolicy.Fixed)
gui.comboBox(value="attr_x", **combo_args)
gui.widgetLabel(self.attr_box, "\u2715", sizePolicy=fixed_size)
gui.comboBox(value="attr_y", **combo_args)
self.vizrank, self.vizrank_button = SieveRank.add_vizrank(
self.attr_box, self, "Score Combinations", self.set_attr)
self.vizrank_button.setSizePolicy(*fixed_size)
self.canvas = QGraphicsScene()
self.canvasView = ViewWithPress(
self.canvas, self.mainArea, handler=self.reset_selection)
self.mainArea.layout().addWidget(self.canvasView)
self.canvasView.setVerticalScrollBarPolicy(Qt.ScrollBarAlwaysOff)
self.canvasView.setHorizontalScrollBarPolicy(Qt.ScrollBarAlwaysOff)
def sizeHint(self):
return QSize(450, 550)
def resizeEvent(self, event):
super().resizeEvent(event)
self.update_graph()
def showEvent(self, event):
super().showEvent(event)
self.update_graph()
@classmethod
def migrate_context(cls, context, version):
if not version:
settings.rename_setting(context, "attrX", "attr_x")
settings.rename_setting(context, "attrY", "attr_y")
settings.migrate_str_to_variable(context)
@Inputs.data
def set_data(self, data):
"""
Discretize continuous attributes, and put all attributes and discrete
metas into self.attrs.
Select the first two attributes unless context overrides this.
Method `resolve_shown_attributes` is called to use the attributes from
the input, if it exists and matches the attributes in the data.
Remove selection; again let the context override this.
Initialize the vizrank dialog, but don't show it.
Args:
data (Table): input data
"""
if isinstance(data, SqlTable) and data.approx_len() > LARGE_TABLE:
data = data.sample_time(DEFAULT_SAMPLE_TIME)
self.closeContext()
self.data = data
self.areas = []
self.selection = set()
if self.data is None:
self.attrs[:] = []
self.domain_model.set_domain(None)
self.discrete_data = None
else:
self.domain_model.set_domain(data.domain)
self.attrs = [x for x in self.domain_model if isinstance(x, Variable)]
if self.attrs:
self.attr_x = self.attrs[0]
self.attr_y = self.attrs[len(self.attrs) > 1]
else:
self.attr_x = self.attr_y = None
self.areas = []
self.selection = set()
self.openContext(self.data)
if self.data:
self.discrete_data = self.sparse_to_dense(data, True)
self.resolve_shown_attributes()
self.update_graph()
self.update_selection()
self.vizrank.initialize()
self.vizrank_button.setEnabled(
self.data is not None and len(self.data) > 1 and
len(self.data.domain.attributes) > 1 and not self.data.is_sparse())
def set_attr(self, attr_x, attr_y):
self.attr_x, self.attr_y = attr_x, attr_y
self.update_attr()
def update_attr(self):
"""Update the graph and selection."""
self.selection = set()
self.discrete_data = self.sparse_to_dense(self.data)
self.update_graph()
self.update_selection()
def sparse_to_dense(self, data, init=False):
"""
Extracts two selected columns from sparse matrix.
GH-2260
"""
def discretizer(data):
if any(attr.is_continuous for attr in chain(data.domain.variables, data.domain.metas)):
discretize = Discretize(
method=EqualFreq(n=4), remove_const=False,
discretize_classes=True, discretize_metas=True)
return discretize(data).to_dense()
return data
if not data.is_sparse() and not init:
return self.discrete_data
if data.is_sparse():
attrs = {self.attr_x,
self.attr_y}
new_domain = data.domain.select_columns(attrs)
data = Table.from_table(new_domain, data)
return discretizer(data)
@Inputs.features
def set_input_features(self, attr_list):
"""
Handler for the Features signal.
The method stores the attributes and calls `resolve_shown_attributes`
Args:
attr_list (AttributeList): data from the signal
"""
self.input_features = attr_list
self.resolve_shown_attributes()
self.update_selection()
def resolve_shown_attributes(self):
"""
Use the attributes from the input signal if the signal is present
and at least two attributes appear in the domain. If there are
multiple, use the first two. Combos are disabled if inputs are used.
"""
self.warning()
self.attr_box.setEnabled(True)
if not self.input_features: # None or empty
return
features = [f for f in self.input_features if f in self.domain_model]
if not features:
self.warning(
"Features from the input signal are not present in the data")
return
old_attrs = self.attr_x, self.attr_y
self.attr_x, self.attr_y = [f for f in (features * 2)[:2]]
self.attr_box.setEnabled(False)
if (self.attr_x, self.attr_y) != old_attrs:
self.selection = set()
self.update_graph()
def reset_selection(self):
self.selection = set()
self.update_selection()
def select_area(self, area, event):
"""
Add or remove the clicked area from the selection
Args:
area (QRect): the area that is clicked
event (QEvent): event description
"""
if event.button() != Qt.LeftButton:
return
index = self.areas.index(area)
if event.modifiers() & Qt.ControlModifier:
self.selection ^= {index}
else:
self.selection = {index}
self.update_selection()
def update_selection(self):
"""
Update the graph (pen width) to show the current selection.
Filter and output the data.
"""
if self.areas is None or not self.selection:
self.Outputs.selected_data.send(None)
self.Outputs.annotated_data.send(create_annotated_table(self.data, []))
return
filts = []
for i, area in enumerate(self.areas):
if i in self.selection:
width = 4
val_x, val_y = area.value_pair
filts.append(
filter.Values([
filter.FilterDiscrete(self.attr_x.name, [val_x]),
filter.FilterDiscrete(self.attr_y.name, [val_y])
]))
else:
width = 1
pen = area.pen()
pen.setWidth(width)
area.setPen(pen)
if len(filts) == 1:
filts = filts[0]
else:
filts = filter.Values(filts, conjunction=False)
selection = filts(self.discrete_data)
idset = set(selection.ids)
sel_idx = [i for i, id in enumerate(self.data.ids) if id in idset]
if self.discrete_data is not self.data:
selection = self.data[sel_idx]
self.Outputs.selected_data.send(selection)
self.Outputs.annotated_data.send(create_annotated_table(self.data, sel_idx))
def update_graph(self):
# Function uses weird names like r, g, b, but it does it with utmost
# caution, hence
# pylint: disable=invalid-name
"""Update the graph."""
def text(txt, *args, **kwargs):
return CanvasText(self.canvas, "", html_text=to_html(txt),
*args, **kwargs)
def width(txt):
return text(txt, 0, 0, show=False).boundingRect().width()
def fmt(val):
return str(int(val)) if val % 1 == 0 else "{:.2f}".format(val)
def show_pearson(rect, pearson, pen_width):
"""
Color the given rectangle according to its corresponding
standardized Pearson residual.
Args:
rect (QRect): the rectangle being drawn
pearson (float): signed standardized pearson residual
pen_width (int): pen width (bolder pen is used for selection)
"""
r = rect.rect()
x, y, w, h = r.x(), r.y(), r.width(), r.height()
if w == 0 or h == 0:
return
r = b = 255
if pearson > 0:
r = g = max(255 - 20 * pearson, 55)
elif pearson < 0:
b = g = max(255 + 20 * pearson, 55)
else:
r = g = b = 224
rect.setBrush(QBrush(QColor(r, g, b)))
pen_color = QColor(255 * (r == 255), 255 * (g == 255),
255 * (b == 255))
pen = QPen(pen_color, pen_width)
rect.setPen(pen)
if pearson > 0:
pearson = min(pearson, 10)
dist = 20 - 1.6 * pearson
else:
pearson = max(pearson, -10)
dist = 20 - 8 * pearson
pen.setWidth(1)
def _offseted_line(ax, ay):
r = QGraphicsLineItem(x + ax, y + ay, x + (ax or w),
y + (ay or h))
self.canvas.addItem(r)
r.setPen(pen)
ax = dist
while ax < w:
_offseted_line(ax, 0)
ax += dist
ay = dist
while ay < h:
_offseted_line(0, ay)
ay += dist
def make_tooltip():
"""Create the tooltip. The function uses local variables from
the enclosing scope."""
# pylint: disable=undefined-loop-variable
def _oper(attr, txt):
if self.data.domain[attr.name] is ddomain[attr.name]:
return "="
return " " if txt[0] in "<≥" else " in "
return (
"<b>{attr_x}{xeq}{xval_name}</b>: {obs_x}/{n} ({p_x:.0f} %)".
format(attr_x=to_html(attr_x.name),
xeq=_oper(attr_x, xval_name),
xval_name=to_html(xval_name),
obs_x=fmt(chi.probs_x[x] * n),
n=int(n),
p_x=100 * chi.probs_x[x]) +
"<br/>" +
"<b>{attr_y}{yeq}{yval_name}</b>: {obs_y}/{n} ({p_y:.0f} %)".
format(attr_y=to_html(attr_y.name),
yeq=_oper(attr_y, yval_name),
yval_name=to_html(yval_name),
obs_y=fmt(chi.probs_y[y] * n),
n=int(n),
p_y=100 * chi.probs_y[y]) +
"<hr/>" +
"""<b>combination of values: </b><br/>
expected {exp} ({p_exp:.0f} %)<br/>
observed {obs} ({p_obs:.0f} %)""".
format(exp=fmt(chi.expected[y, x]),
p_exp=100 * chi.expected[y, x] / n,
obs=fmt(chi.observed[y, x]),
p_obs=100 * chi.observed[y, x] / n))
for item in self.canvas.items():
self.canvas.removeItem(item)
if self.data is None or len(self.data) == 0 or \
self.attr_x is None or self.attr_y is None:
return
ddomain = self.discrete_data.domain
attr_x, attr_y = self.attr_x, self.attr_y
disc_x, disc_y = ddomain[attr_x.name], ddomain[attr_y.name]
view = self.canvasView
chi = ChiSqStats(self.discrete_data, disc_x, disc_y)
max_ylabel_w = max((width(val) for val in disc_y.values), default=0)
max_ylabel_w = min(max_ylabel_w, 200)
x_off = width(attr_x.name) + max_ylabel_w
y_off = 15
square_size = min(view.width() - x_off - 35, view.height() - y_off - 80)
square_size = max(square_size, 10)
self.canvasView.setSceneRect(0, 0, view.width(), view.height())
if not disc_x.values or not disc_y.values:
text_ = "Features {} and {} have no values".format(disc_x, disc_y) \
if not disc_x.values and \
not disc_y.values and \
disc_x != disc_y \
else \
"Feature {} has no values".format(
disc_x if not disc_x.values else disc_y)
text(text_, view.width() / 2 + 70, view.height() / 2,
Qt.AlignRight | Qt.AlignVCenter)
return
n = chi.n
curr_x = x_off
max_xlabel_h = 0
self.areas = []
for x, (px, xval_name) in enumerate(zip(chi.probs_x, disc_x.values)):
if px == 0:
continue
width = square_size * px
curr_y = y_off
for y in range(len(chi.probs_y) - 1, -1, -1): # bottom-up order
py = chi.probs_y[y]
yval_name = disc_y.values[y]
if py == 0:
continue
height = square_size * py
selected = len(self.areas) in self.selection
rect = CanvasRectangle(
self.canvas, curr_x + 2, curr_y + 2, width - 4, height - 4,
z=-10, onclick=self.select_area)
rect.value_pair = x, y
self.areas.append(rect)
show_pearson(rect, chi.residuals[y, x], 3 * selected)
rect.setToolTip(make_tooltip())
if x == 0:
text(yval_name, x_off, curr_y + height / 2,
Qt.AlignRight | Qt.AlignVCenter)
curr_y += height
xl = text(xval_name, curr_x + width / 2, y_off + square_size,
Qt.AlignHCenter | Qt.AlignTop)
max_xlabel_h = max(int(xl.boundingRect().height()), max_xlabel_h)
curr_x += width
bottom = y_off + square_size + max_xlabel_h
text(attr_y.name, 0, y_off + square_size / 2,
Qt.AlignLeft | Qt.AlignVCenter, bold=True, vertical=True)
text(attr_x.name, x_off + square_size / 2, bottom,
Qt.AlignHCenter | Qt.AlignTop, bold=True)
bottom += 30
xl = text("χ²={:.2f}, p={:.3f}".format(chi.chisq, chi.p),
0, bottom)
# Assume similar height for both lines
text("N = " + fmt(chi.n), 0, bottom - xl.boundingRect().height())
def get_widget_name_extension(self):
if self.data is not None:
return "{} vs {}".format(self.attr_x.name, self.attr_y.name)
def send_report(self):
self.report_plot()
class OWExplainPrediction(OWWidget, ConcurrentWidgetMixin):
name = "Explain Prediction"
description = "Prediction explanation widget."
icon = "icons/ExplainPred.svg"
priority = 110
class Inputs:
model = Input("Model", Model)
background_data = Input("Background Data", Table)
data = Input("Data", Table)
class Outputs:
scores = Output("Scores", Table)
class Error(OWWidget.Error):
domain_transform_err = Msg("{}")
unknown_err = Msg("{}")
class Information(OWWidget.Information):
multiple_instances = Msg("Explaining prediction for the first "
"instance in 'Data'.")
settingsHandler = ClassValuesContextHandler()
target_index = ContextSetting(0)
stripe_len = Setting(10)
graph_name = "scene"
def __init__(self):
OWWidget.__init__(self)
ConcurrentWidgetMixin.__init__(self)
self.__results = None # type: Optional[Results]
self.model = None # type: Optional[Model]
self.background_data = None # type: Optional[Table]
self.data = None # type: Optional[Table]
self._stripe_plot = None # type: Optional[StripePlot]
self.mo_info = ""
self.bv_info = ""
self.setup_gui()
def setup_gui(self):
self._add_controls()
self._add_plot()
self.info.set_input_summary(self.info.NoInput)
def _add_plot(self):
self.scene = QGraphicsScene()
self.view = QGraphicsView(self.scene)
self.view.setRenderHint(QPainter.Antialiasing, True)
self.view.setAlignment(Qt.AlignVCenter | Qt.AlignLeft)
self.mainArea.layout().addWidget(self.view)
def _add_controls(self):
box = gui.vBox(self.controlArea, "Target class")
self._target_combo = gui.comboBox(box,
self,
"target_index",
callback=self.__target_combo_changed,
contentsLength=12)
box = gui.hBox(self.controlArea, "Zoom")
gui.hSlider(box,
self,
"stripe_len",
None,
minValue=1,
maxValue=500,
createLabel=False,
callback=self.__size_slider_changed)
gui.rubber(self.controlArea)
box = gui.vBox(self.controlArea, "Prediction info")
gui.label(box, self, "%(mo_info)s") # type: QLabel
bv_label = gui.label(box, self, "%(bv_info)s") # type: QLabel
bv_label.setToolTip("The average prediction for selected class.")
def __target_combo_changed(self):
self.update_scene()
def __size_slider_changed(self):
if self._stripe_plot is not None:
self._stripe_plot.set_height(self.stripe_len)
@Inputs.data
@check_sql_input
def set_data(self, data: Optional[Table]):
self.data = data
@Inputs.background_data
@check_sql_input
def set_background_data(self, data: Optional[Table]):
self.background_data = data
@Inputs.model
def set_model(self, model: Optional[Model]):
self.closeContext()
self.model = model
self.setup_controls()
self.openContext(self.model.domain.class_var if self.model else None)
def setup_controls(self):
self._target_combo.clear()
self._target_combo.setEnabled(True)
if self.model is not None:
if self.model.domain.has_discrete_class:
self._target_combo.addItems(self.model.domain.class_var.values)
self.target_index = 0
elif self.model.domain.has_continuous_class:
self.target_index = -1
self._target_combo.setEnabled(False)
else:
raise NotImplementedError
def handleNewSignals(self):
self.clear()
self.check_inputs()
data = self.data and self.data[:1]
self.start(run, data, self.background_data, self.model)
def clear(self):
self.mo_info = ""
self.bv_info = ""
self.__results = None
self.cancel()
self.clear_scene()
self.clear_messages()
def check_inputs(self):
if self.data and len(self.data) > 1:
self.Information.multiple_instances()
summary, details, kwargs = self.info.NoInput, "", {}
if self.data or self.background_data:
n_data = len(self.data) if self.data else 0
n_background_data = len(self.background_data) \
if self.background_data else 0
summary = f"{self.info.format_number(n_background_data)}, " \
f"{self.info.format_number(n_data)}"
kwargs = {"format": Qt.RichText}
details = format_multiple_summaries([("Background data",
self.background_data),
("Data", self.data)])
self.info.set_input_summary(summary, details, **kwargs)
def clear_scene(self):
self.scene.clear()
self.scene.setSceneRect(QRectF())
self.view.setSceneRect(QRectF())
self._stripe_plot = None
def update_scene(self):
self.clear_scene()
self.mo_info = ""
self.bv_info = ""
scores = None
if self.__results is not None:
data = self.__results.transformed_data
pred = self.__results.predictions
base = self.__results.base_value
values, _, labels, ranges = prepare_force_plot_data(
self.__results.values, data, pred, self.target_index)
index = 0
HIGH, LOW = 0, 1
plot_data = PlotData(high_values=values[index][HIGH],
low_values=values[index][LOW][::-1],
high_labels=labels[index][HIGH],
low_labels=labels[index][LOW][::-1],
value_range=ranges[index],
model_output=pred[index][self.target_index],
base_value=base[self.target_index])
self.setup_plot(plot_data)
self.mo_info = f"Model prediction: {_str(plot_data.model_output)}"
self.bv_info = f"Base value: {_str(plot_data.base_value)}"
assert isinstance(self.__results.values, list)
scores = self.__results.values[self.target_index][0, :]
names = [a.name for a in data.domain.attributes]
scores = self.create_scores_table(scores, names)
self.Outputs.scores.send(scores)
def setup_plot(self, plot_data: PlotData):
self._stripe_plot = StripePlot()
self._stripe_plot.set_data(plot_data, self.stripe_len)
self._stripe_plot.layout().activate()
self._stripe_plot.geometryChanged.connect(self.update_scene_rect)
self.scene.addItem(self._stripe_plot)
self.update_scene_rect()
def update_scene_rect(self):
geom = self._stripe_plot.geometry()
self.scene.setSceneRect(geom)
self.view.setSceneRect(geom)
@staticmethod
def create_scores_table(scores: np.ndarray, names: List[str]) -> Table:
domain = Domain([ContinuousVariable("Score")],
metas=[StringVariable("Feature")])
scores_table = Table(domain,
scores[:, None],
metas=np.array(names)[:, None])
scores_table.name = "Feature Scores"
return scores_table
def on_partial_result(self, _):
pass
def on_done(self, results: Optional[RunnerResults]):
self.__results = results
self.update_scene()
def on_exception(self, ex: Exception):
if isinstance(ex, DomainTransformationError):
self.Error.domain_transform_err(ex)
else:
self.Error.unknown_err(ex)
def onDeleteWidget(self):
self.shutdown()
super().onDeleteWidget()
def sizeHint(self) -> QSizeF:
sh = self.controlArea.sizeHint()
return sh.expandedTo(QSize(700, 700))
def send_report(self):
if not self.data or not self.background_data or not self.model:
return
items = {"Target class": "None"}
if self.model.domain.has_discrete_class:
class_var = self.model.domain.class_var
items["Target class"] = class_var.values[self.target_index]
self.report_items(items)
self.report_plot()
class OWMosaicDisplay(OWWidget):
name = "Mosaic Display"
description = "Display data in a mosaic plot."
icon = "icons/MosaicDisplay.svg"
priority = 220
inputs = [("Data", Table, "set_data", Default),
("Data Subset", Table, "set_subset_data")]
outputs = [("Selected Data", Table, widget.Default),
(ANNOTATED_DATA_SIGNAL_NAME, Table)]
settingsHandler = DomainContextHandler()
use_boxes = Setting(True)
variable1 = ContextSetting("", exclude_metas=False)
variable2 = ContextSetting("", exclude_metas=False)
variable3 = ContextSetting("", exclude_metas=False)
variable4 = ContextSetting("", exclude_metas=False)
selection = ContextSetting(set())
# interior_coloring is context setting to properly reset it
# if the widget switches to regression and back (set setData)
interior_coloring = ContextSetting(1)
PEARSON, CLASS_DISTRIBUTION = 0, 1
interior_coloring_opts = ["Pearson residuals",
"Class distribution"]
BAR_WIDTH = 5
SPACING = 4
ATTR_NAME_OFFSET = 20
ATTR_VAL_OFFSET = 3
BLUE_COLORS = [QColor(255, 255, 255), QColor(210, 210, 255),
QColor(110, 110, 255), QColor(0, 0, 255)]
RED_COLORS = [QColor(255, 255, 255), QColor(255, 200, 200),
QColor(255, 100, 100), QColor(255, 0, 0)]
graph_name = "canvas"
class Warning(OWWidget.Warning):
incompatible_subset = Msg("Data subset is incompatible with Data")
no_valid_data = Msg("No valid data")
no_cont_selection_sql = \
Msg("Selection of continuous variables on SQL is not supported")
def __init__(self):
super().__init__()
self.data = None
self.discrete_data = None
self.unprocessed_subset_data = None
self.subset_data = None
self.areas = []
self.canvas = QGraphicsScene()
self.canvas_view = ViewWithPress(self.canvas,
handler=self.clear_selection)
self.mainArea.layout().addWidget(self.canvas_view)
self.canvas_view.setVerticalScrollBarPolicy(Qt.ScrollBarAlwaysOff)
self.canvas_view.setHorizontalScrollBarPolicy(Qt.ScrollBarAlwaysOff)
self.canvas_view.setRenderHint(QPainter.Antialiasing)
box = gui.vBox(self.controlArea, box=True)
self.attr_combos = [
gui.comboBox(
box, self, value="variable{}".format(i),
orientation=Qt.Horizontal, contentsLength=12,
callback=self.reset_graph,
sendSelectedValue=True, valueType=str)
for i in range(1, 5)]
self.rb_colors = gui.radioButtonsInBox(
self.controlArea, self, "interior_coloring",
self.interior_coloring_opts, box="Interior Coloring",
callback=self.update_graph)
self.bar_button = gui.checkBox(
gui.indentedBox(self.rb_colors),
self, 'use_boxes', label='Compare with total',
callback=self._compare_with_total)
gui.rubber(self.controlArea)
def sizeHint(self):
return QSize(530, 720)
def _compare_with_total(self):
if self.data and self.data.domain.has_discrete_class:
self.interior_coloring = 1
self.update_graph()
def init_combos(self, data):
for combo in self.attr_combos:
combo.clear()
if data is None:
return
for combo in self.attr_combos[1:]:
combo.addItem("(None)")
icons = gui.attributeIconDict
for attr in chain(data.domain, data.domain.metas):
if attr.is_discrete or attr.is_continuous:
for combo in self.attr_combos:
combo.addItem(icons[attr], attr.name)
if self.attr_combos[0].count() > 0:
self.variable1 = self.attr_combos[0].itemText(0)
self.variable2 = self.attr_combos[1].itemText(
2 * (self.attr_combos[1].count() > 2))
self.variable3 = self.attr_combos[2].itemText(0)
self.variable4 = self.attr_combos[3].itemText(0)
def get_attr_list(self):
return [
a for a in [self.variable1, self.variable2,
self.variable3, self.variable4]
if a and a != "(None)"]
def resizeEvent(self, e):
OWWidget.resizeEvent(self, e)
self.update_graph()
def showEvent(self, ev):
OWWidget.showEvent(self, ev)
self.update_graph()
def set_data(self, data):
if type(data) == SqlTable and data.approx_len() > LARGE_TABLE:
data = data.sample_time(DEFAULT_SAMPLE_TIME)
self.closeContext()
self.data = data
self.init_combos(self.data)
if not self.data:
self.discrete_data = None
return
if any(attr.is_continuous for attr in data.domain):
self.discrete_data = Discretize(method=EqualFreq(n=4))(data)
else:
self.discrete_data = self.data
if self.data.domain.class_var is None:
self.rb_colors.setDisabled(True)
disc_class = False
else:
self.rb_colors.setDisabled(False)
disc_class = self.data.domain.has_discrete_class
self.rb_colors.group.button(2).setDisabled(not disc_class)
self.bar_button.setDisabled(not disc_class)
self.interior_coloring = bool(disc_class)
self.openContext(self.data)
# if we first received subset we now call setSubsetData to process it
if self.unprocessed_subset_data:
self.set_subset_data(self.unprocessed_subset_data)
self.unprocessed_subset_data = None
def set_subset_data(self, data):
self.Warning.incompatible_subset.clear()
if self.data is None:
self.unprocessed_subset_data = data
return
try:
self.subset_data = data.from_table(self.data.domain, data)
except:
self.subset_data = None
self.Warning.incompatible_subset(shown=data is not None)
# this is called by widget after setData and setSubsetData are called.
# this way the graph is updated only once
def handleNewSignals(self):
self.reset_graph()
def clear_selection(self):
self.selection = set()
self.update_selection_rects()
self.send_selection()
def reset_graph(self):
self.clear_selection()
self.update_graph()
def update_selection_rects(self):
for i, (attr, vals, area) in enumerate(self.areas):
if i in self.selection:
area.setPen(QPen(Qt.black, 3, Qt.DotLine))
else:
area.setPen(QPen())
def select_area(self, index, ev):
if ev.button() != Qt.LeftButton:
return
if ev.modifiers() & Qt.ControlModifier:
self.selection ^= {index}
else:
self.selection = {index}
self.update_selection_rects()
self.send_selection()
def send_selection(self):
if not self.selection or self.data is None:
self.send("Selected Data", None)
self.send(ANNOTATED_DATA_SIGNAL_NAME,
create_annotated_table(self.data, []))
return
filters = []
self.Warning.no_cont_selection_sql.clear()
if self.discrete_data is not self.data:
if isinstance(self.data, SqlTable):
self.Warning.no_cont_selection_sql()
for i in self.selection:
cols, vals, area = self.areas[i]
filters.append(
filter.Values(
filter.FilterDiscrete(col, [val])
for col, val in zip(cols, vals)))
if len(filters) > 1:
filters = filter.Values(filters, conjunction=False)
else:
filters = filters[0]
selection = filters(self.discrete_data)
idset = set(selection.ids)
sel_idx = [i for i, id in enumerate(self.data.ids) if id in idset]
if self.discrete_data is not self.data:
selection = self.data[sel_idx]
self.send("Selected Data", selection)
self.send(ANNOTATED_DATA_SIGNAL_NAME,
create_annotated_table(self.data, sel_idx))
def send_report(self):
self.report_plot(self.canvas)
def update_graph(self):
spacing = self.SPACING
bar_width = self.BAR_WIDTH
def draw_data(attr_list, x0_x1, y0_y1, side, condition,
total_attrs, used_attrs=[], used_vals=[],
attr_vals=""):
x0, x1 = x0_x1
y0, y1 = y0_y1
if conditionaldict[attr_vals] == 0:
add_rect(x0, x1, y0, y1, "",
used_attrs, used_vals, attr_vals=attr_vals)
# store coordinates for later drawing of labels
draw_text(side, attr_list[0], (x0, x1), (y0, y1), total_attrs,
used_attrs, used_vals, attr_vals)
return
attr = attr_list[0]
# how much smaller rectangles do we draw
edge = len(attr_list) * spacing
values = get_variable_values_sorted(data.domain[attr])
if side % 2:
values = values[::-1] # reverse names if necessary
if side % 2 == 0: # we are drawing on the x axis
# remove the space needed for separating different attr. values
whole = max(0, (x1 - x0) - edge * (
len(values) - 1))
if whole == 0:
edge = (x1 - x0) / float(len(values) - 1)
else: # we are drawing on the y axis
whole = max(0, (y1 - y0) - edge * (len(values) - 1))
if whole == 0:
edge = (y1 - y0) / float(len(values) - 1)
if attr_vals == "":
counts = [conditionaldict[val] for val in values]
else:
counts = [conditionaldict[attr_vals + "-" + val]
for val in values]
total = sum(counts)
# if we are visualizing the third attribute and the first attribute
# has the last value, we have to reverse the order in which the
# boxes will be drawn otherwise, if the last cell, nearest to the
# labels of the fourth attribute, is empty, we wouldn't be able to
# position the labels
valrange = list(range(len(values)))
if len(attr_list + used_attrs) == 4 and len(used_attrs) == 2:
attr1values = get_variable_values_sorted(
data.domain[used_attrs[0]])
if used_vals[0] == attr1values[-1]:
valrange = valrange[::-1]
for i in valrange:
start = i * edge + whole * float(sum(counts[:i]) / total)
end = i * edge + whole * float(sum(counts[:i + 1]) / total)
val = values[i]
htmlval = to_html(val)
if attr_vals != "":
newattrvals = attr_vals + "-" + val
else:
newattrvals = val
tooltip = condition + 4 * " " + attr + \
": <b>" + htmlval + "</b><br>"
attrs = used_attrs + [attr]
vals = used_vals + [val]
common_args = attrs, vals, newattrvals
if side % 2 == 0: # if we are moving horizontally
if len(attr_list) == 1:
add_rect(x0 + start, x0 + end, y0, y1,
tooltip, *common_args)
else:
draw_data(attr_list[1:], (x0 + start, x0 + end),
(y0, y1), side + 1,
tooltip, total_attrs, *common_args)
else:
if len(attr_list) == 1:
add_rect(x0, x1, y0 + start, y0 + end,
tooltip, *common_args)
else:
draw_data(attr_list[1:], (x0, x1),
(y0 + start, y0 + end), side + 1,
tooltip, total_attrs, *common_args)
draw_text(side, attr_list[0], (x0, x1), (y0, y1),
total_attrs, used_attrs, used_vals, attr_vals)
def draw_text(side, attr, x0_x1, y0_y1,
total_attrs, used_attrs, used_vals, attr_vals):
x0, x1 = x0_x1
y0, y1 = y0_y1
if side in drawn_sides:
return
# the text on the right will be drawn when we are processing
# visualization of the last value of the first attribute
if side == 3:
attr1values = \
get_variable_values_sorted(data.domain[used_attrs[0]])
if used_vals[0] != attr1values[-1]:
return
if not conditionaldict[attr_vals]:
if side not in draw_positions:
draw_positions[side] = (x0, x1, y0, y1)
return
else:
if side in draw_positions:
# restore the positions of attribute values and name
(x0, x1, y0, y1) = draw_positions[side]
drawn_sides.add(side)
values = get_variable_values_sorted(data.domain[attr])
if side % 2:
values = values[::-1]
spaces = spacing * (total_attrs - side) * (len(values) - 1)
width = x1 - x0 - spaces * (side % 2 == 0)
height = y1 - y0 - spaces * (side % 2 == 1)
# calculate position of first attribute
currpos = 0
if attr_vals == "":
counts = [conditionaldict.get(val, 1) for val in values]
else:
counts = [conditionaldict.get(attr_vals + "-" + val, 1)
for val in values]
total = sum(counts)
if total == 0:
counts = [1] * len(values)
total = sum(counts)
aligns = [Qt.AlignTop | Qt.AlignHCenter,
Qt.AlignRight | Qt.AlignVCenter,
Qt.AlignBottom | Qt.AlignHCenter,
Qt.AlignLeft | Qt.AlignVCenter]
align = aligns[side]
for i in range(len(values)):
val = values[i]
perc = counts[i] / float(total)
if distributiondict[val] != 0:
if side == 0:
CanvasText(self.canvas, str(val),
x0 + currpos + width * 0.5 * perc,
y1 + self.ATTR_VAL_OFFSET, align)
elif side == 1:
CanvasText(self.canvas, str(val),
x0 - self.ATTR_VAL_OFFSET,
y0 + currpos + height * 0.5 * perc, align)
elif side == 2:
CanvasText(self.canvas, str(val),
x0 + currpos + width * perc * 0.5,
y0 - self.ATTR_VAL_OFFSET, align)
else:
CanvasText(self.canvas, str(val),
x1 + self.ATTR_VAL_OFFSET,
y0 + currpos + height * 0.5 * perc, align)
if side % 2 == 0:
currpos += perc * width + spacing * (total_attrs - side)
else:
currpos += perc * height + spacing * (total_attrs - side)
if side == 0:
CanvasText(
self.canvas, attr,
x0 + (x1 - x0) / 2,
y1 + self.ATTR_VAL_OFFSET +
self.ATTR_NAME_OFFSET,
align, bold=1)
elif side == 1:
CanvasText(
self.canvas, attr,
x0 - max_ylabel_w1 - self.ATTR_VAL_OFFSET,
y0 + (y1 - y0) / 2,
align, bold=1, vertical=True)
elif side == 2:
CanvasText(
self.canvas, attr,
x0 + (x1 - x0) / 2,
y0 - self.ATTR_VAL_OFFSET -
self.ATTR_NAME_OFFSET,
align, bold=1)
else:
CanvasText(
self.canvas, attr,
x1 + max_ylabel_w2 + self.ATTR_VAL_OFFSET,
y0 + (y1 - y0) / 2,
align, bold=1, vertical=True)
def add_rect(x0, x1, y0, y1, condition="",
used_attrs=[], used_vals=[], attr_vals=""):
area_index = len(self.areas)
if x0 == x1:
x1 += 1
if y0 == y1:
y1 += 1
# rectangles of width and height 1 are not shown - increase
if x1 - x0 + y1 - y0 == 2:
y1 += 1
if class_var and class_var.is_discrete:
colors = [QColor(*col) for col in class_var.colors]
else:
colors = None
def select_area(_, ev):
self.select_area(area_index, ev)
def rect(x, y, w, h, z, pen_color=None, brush_color=None, **args):
if pen_color is None:
return CanvasRectangle(
self.canvas, x, y, w, h, z=z, onclick=select_area,
**args)
if brush_color is None:
brush_color = pen_color
return CanvasRectangle(
self.canvas, x, y, w, h, pen_color, brush_color, z=z,
onclick=select_area, **args)
def line(x1, y1, x2, y2):
r = QGraphicsLineItem(x1, y1, x2, y2, None)
self.canvas.addItem(r)
r.setPen(QPen(Qt.white, 2))
r.setZValue(30)
outer_rect = rect(x0, y0, x1 - x0, y1 - y0, 30)
self.areas.append((used_attrs, used_vals, outer_rect))
if not conditionaldict[attr_vals]:
return
if self.interior_coloring == self.PEARSON:
s = sum(apriori_dists[0])
expected = s * reduce(
mul,
(apriori_dists[i][used_vals[i]] / float(s)
for i in range(len(used_vals))))
actual = conditionaldict[attr_vals]
pearson = (actual - expected) / sqrt(expected)
if pearson == 0:
ind = 0
else:
ind = max(0, min(int(log(abs(pearson), 2)), 3))
color = [self.RED_COLORS, self.BLUE_COLORS][pearson > 0][ind]
rect(x0, y0, x1 - x0, y1 - y0, -20, color)
outer_rect.setToolTip(
condition + "<hr/>" +
"Expected instances: %.1f<br>"
"Actual instances: %d<br>"
"Standardized (Pearson) residual: %.1f" %
(expected, conditionaldict[attr_vals], pearson))
else:
cls_values = get_variable_values_sorted(class_var)
prior = get_distribution(data, class_var.name)
total = 0
for i, value in enumerate(cls_values):
val = conditionaldict[attr_vals + "-" + value]
if val == 0:
continue
if i == len(cls_values) - 1:
v = y1 - y0 - total
else:
v = ((y1 - y0) * val) / conditionaldict[attr_vals]
rect(x0, y0 + total, x1 - x0, v, -20, colors[i])
total += v
if self.use_boxes and \
abs(x1 - x0) > bar_width and \
abs(y1 - y0) > bar_width:
total = 0
line(x0 + bar_width, y0, x0 + bar_width, y1)
n = sum(prior)
for i, (val, color) in enumerate(zip(prior, colors)):
if i == len(prior) - 1:
h = y1 - y0 - total
else:
h = (y1 - y0) * val / n
rect(x0, y0 + total, bar_width, h, 20, color)
total += h
if conditionalsubsetdict:
if conditionalsubsetdict[attr_vals]:
counts = [conditionalsubsetdict[attr_vals + "-" + val]
for val in cls_values]
if sum(counts) == 1:
rect(x0 - 2, y0 - 2, x1 - x0 + 5, y1 - y0 + 5, -550,
colors[counts.index(1)], Qt.white,
penWidth=2, penStyle=Qt.DashLine)
if self.subset_data is not None:
line(x1 - bar_width, y0, x1 - bar_width, y1)
total = 0
n = conditionalsubsetdict[attr_vals]
if n:
for i, (cls, color) in \
enumerate(zip(cls_values, colors)):
val = conditionalsubsetdict[
attr_vals + "-" + cls]
if val == 0:
continue
if i == len(prior) - 1:
v = y1 - y0 - total
else:
v = ((y1 - y0) * val) / n
rect(x1 - bar_width, y0 + total,
bar_width, v, 15, color)
total += v
actual = [conditionaldict[attr_vals + "-" + cls_values[i]]
for i in range(len(prior))]
n_actual = sum(actual)
if n_actual > 0:
apriori = [prior[key] for key in cls_values]
n_apriori = sum(apriori)
text = "<br/>".join(
"<b>%s</b>: %d / %.1f%% (Expected %.1f / %.1f%%)" %
(cls, act, 100.0 * act / n_actual,
apr / n_apriori * n_actual, 100.0 * apr / n_apriori
)
for cls, act, apr in zip(cls_values, actual, apriori
))
else:
text = ""
outer_rect.setToolTip(
"{}<hr>Instances: {}<br><br>{}".format(
condition, n_actual, text[:-4]))
def draw_legend(x0_x1, y0_y1):
x0, x1 = x0_x1
y0, y1 = y0_y1
if self.interior_coloring == self.PEARSON:
names = ["<-8", "-8:-4", "-4:-2", "-2:2", "2:4", "4:8", ">8",
"Residuals:"]
colors = self.RED_COLORS[::-1] + self.BLUE_COLORS[1:]
else:
names = get_variable_values_sorted(class_var) + \
[class_var.name + ":"]
colors = [QColor(*col) for col in class_var.colors]
names = [CanvasText(self.canvas, name, alignment=Qt.AlignVCenter)
for name in names]
totalwidth = sum(text.boundingRect().width() for text in names)
# compute the x position of the center of the legend
y = y1 + self.ATTR_NAME_OFFSET + self.ATTR_VAL_OFFSET + 35
distance = 30
startx = (x0 + x1) / 2 - (totalwidth + (len(names)) * distance) / 2
names[-1].setPos(startx + 15, y)
names[-1].show()
xoffset = names[-1].boundingRect().width() + distance
size = 8
for i in range(len(names) - 1):
if self.interior_coloring == self.PEARSON:
edgecolor = Qt.black
else:
edgecolor = colors[i]
CanvasRectangle(self.canvas, startx + xoffset, y - size / 2,
size, size, edgecolor, colors[i])
names[i].setPos(startx + xoffset + 10, y)
xoffset += distance + names[i].boundingRect().width()
self.canvas.clear()
self.areas = []
data = self.discrete_data
if data is None:
return
subset = self.subset_data
attr_list = self.get_attr_list()
class_var = data.domain.class_var
if class_var:
sql = type(data) == SqlTable
name = not sql and data.name
# save class_var because it is removed in the next line
data = data[:, attr_list + [class_var]]
data.domain.class_var = class_var
if not sql:
data.name = name
else:
data = data[:, attr_list]
# TODO: check this
# data = Preprocessor_dropMissing(data)
if len(data) == 0:
self.Warning.no_valid_data()
return
else:
self.Warning.no_valid_data.clear()
if self.interior_coloring == self.PEARSON:
apriori_dists = [get_distribution(data, attr) for attr in attr_list]
else:
apriori_dists = []
def get_max_label_width(attr):
values = get_variable_values_sorted(data.domain[attr])
maxw = 0
for val in values:
t = CanvasText(self.canvas, val, 0, 0, bold=0, show=False)
maxw = max(int(t.boundingRect().width()), maxw)
return maxw
# get the maximum width of rectangle
xoff = 20
width = 20
if len(attr_list) > 1:
text = CanvasText(self.canvas, attr_list[1], bold=1, show=0)
max_ylabel_w1 = min(get_max_label_width(attr_list[1]), 150)
width = 5 + text.boundingRect().height() + \
self.ATTR_VAL_OFFSET + max_ylabel_w1
xoff = width
if len(attr_list) == 4:
text = CanvasText(self.canvas, attr_list[3], bold=1, show=0)
max_ylabel_w2 = min(get_max_label_width(attr_list[3]), 150)
width += text.boundingRect().height() + \
self.ATTR_VAL_OFFSET + max_ylabel_w2 - 10
# get the maximum height of rectangle
height = 100
yoff = 45
square_size = min(self.canvas_view.width() - width - 20,
self.canvas_view.height() - height - 20)
if square_size < 0:
return # canvas is too small to draw rectangles
self.canvas_view.setSceneRect(
0, 0, self.canvas_view.width(), self.canvas_view.height())
drawn_sides = set()
draw_positions = {}
conditionaldict, distributiondict = \
get_conditional_distribution(data, attr_list)
conditionalsubsetdict = None
if subset:
conditionalsubsetdict, _ = \
get_conditional_distribution(subset, attr_list)
# draw rectangles
draw_data(
attr_list, (xoff, xoff + square_size), (yoff, yoff + square_size),
0, "", len(attr_list))
draw_legend((xoff, xoff + square_size), (yoff, yoff + square_size))
self.update_selection_rects()
class OWPythagoreanForest(OWWidget):
name = 'Pythagorean Forest'
description = 'Pythagorean forest for visualising random forests.'
icon = 'icons/PythagoreanForest.svg'
priority = 1001
inputs = [('Random forest', RandomForestModel, 'set_rf')]
outputs = [('Tree', TreeModel)]
# Enable the save as feature
graph_name = 'scene'
# Settings
depth_limit = settings.ContextSetting(10)
target_class_index = settings.ContextSetting(0)
size_calc_idx = settings.Setting(0)
zoom = settings.Setting(50)
selected_tree_index = settings.ContextSetting(-1)
def __init__(self):
super().__init__()
self.model = None
self.forest_adapter = None
self.instances = None
self.clf_dataset = None
# We need to store refernces to the trees and grid items
self.grid_items, self.ptrees = [], []
# In some rare cases, we need to prevent commiting, the only one
# that this currently helps is that when changing the size calculation
# the trees are all recomputed, but we don't want to output a new tree
# to keep things consistent with other ui controls.
self.__prevent_commit = False
self.color_palette = None
# Different methods to calculate the size of squares
self.SIZE_CALCULATION = [
('Normal', lambda x: x),
('Square root', lambda x: sqrt(x)),
('Logarithmic', lambda x: log(x + 1)),
]
# CONTROL AREA
# Tree info area
box_info = gui.widgetBox(self.controlArea, 'Forest')
self.ui_info = gui.widgetLabel(box_info)
# Display controls area
box_display = gui.widgetBox(self.controlArea, 'Display')
self.ui_depth_slider = gui.hSlider(box_display,
self,
'depth_limit',
label='Depth',
ticks=False,
callback=self.update_depth)
self.ui_target_class_combo = gui.comboBox(box_display,
self,
'target_class_index',
label='Target class',
orientation=Qt.Horizontal,
items=[],
contentsLength=8,
callback=self.update_colors)
self.ui_size_calc_combo = gui.comboBox(
box_display,
self,
'size_calc_idx',
label='Size',
orientation=Qt.Horizontal,
items=list(zip(*self.SIZE_CALCULATION))[0],
contentsLength=8,
callback=self.update_size_calc)
self.ui_zoom_slider = gui.hSlider(box_display,
self,
'zoom',
label='Zoom',
ticks=False,
minValue=20,
maxValue=150,
callback=self.zoom_changed,
createLabel=False)
# Stretch to fit the rest of the unsused area
gui.rubber(self.controlArea)
self.controlArea.setSizePolicy(QSizePolicy.Preferred,
QSizePolicy.Expanding)
# MAIN AREA
self.scene = QGraphicsScene(self)
self.scene.selectionChanged.connect(self.commit)
self.grid = OWGrid()
self.grid.geometryChanged.connect(self._update_scene_rect)
self.scene.addItem(self.grid)
self.view = QGraphicsView(self.scene)
self.view.setRenderHint(QPainter.Antialiasing, True)
self.view.setVerticalScrollBarPolicy(Qt.ScrollBarAlwaysOn)
self.mainArea.layout().addWidget(self.view)
self.resize(800, 500)
self.clear()
def set_rf(self, model=None):
"""When a different forest is given."""
self.clear()
self.model = model
if model is not None:
self.forest_adapter = self._get_forest_adapter(self.model)
self._draw_trees()
self.color_palette = self.forest_adapter.get_trees()[0]
self.instances = model.instances
# this bit is important for the regression classifier
if self.instances is not None and self.instances.domain != model.domain:
self.clf_dataset = Table.from_table(self.model.domain,
self.instances)
else:
self.clf_dataset = self.instances
self._update_info_box()
self._update_target_class_combo()
self._update_depth_slider()
self.selected_tree_index = -1
def clear(self):
"""Clear all relevant data from the widget."""
self.model = None
self.forest_adapter = None
self.ptrees = []
self.grid_items = []
self.grid.clear()
self._clear_info_box()
self._clear_target_class_combo()
self._clear_depth_slider()
def update_depth(self):
"""When the max depth slider is changed."""
for tree in self.ptrees:
tree.set_depth_limit(self.depth_limit)
def update_colors(self):
"""When the target class or coloring method is changed."""
for tree in self.ptrees:
tree.target_class_changed(self.target_class_index)
def update_size_calc(self):
"""When the size calculation of the trees is changed."""
if self.model is not None:
with self._prevent_commit():
self.grid.clear()
self._draw_trees()
# Keep the selected item
if self.selected_tree_index != -1:
self.grid_items[self.selected_tree_index].setSelected(True)
self.update_depth()
def zoom_changed(self):
"""When we update the "Zoom" slider."""
for item in self.grid_items:
item.set_max_size(self._calculate_zoom(self.zoom))
width = (self.view.width() - self.view.verticalScrollBar().width())
self.grid.reflow(width)
self.grid.setPreferredWidth(width)
@contextmanager
def _prevent_commit(self):
try:
self.__prevent_commit = True
yield
finally:
self.__prevent_commit = False
def _update_info_box(self):
self.ui_info.setText('Trees: {}'.format(
len(self.forest_adapter.get_trees())))
def _update_depth_slider(self):
self.depth_limit = self._get_max_depth()
self.ui_depth_slider.parent().setEnabled(True)
self.ui_depth_slider.setMaximum(self.depth_limit)
self.ui_depth_slider.setValue(self.depth_limit)
def _clear_info_box(self):
self.ui_info.setText('No forest on input.')
def _clear_target_class_combo(self):
self.ui_target_class_combo.clear()
self.target_class_index = 0
self.ui_target_class_combo.setCurrentIndex(self.target_class_index)
def _clear_depth_slider(self):
self.ui_depth_slider.parent().setEnabled(False)
self.ui_depth_slider.setMaximum(0)
def _get_max_depth(self):
return max(tree.tree_adapter.max_depth for tree in self.ptrees)
def _get_forest_adapter(self, model):
return SklRandomForestAdapter(model)
@contextmanager
def disable_ui(self):
"""Temporarly disable the UI while trees may be redrawn."""
try:
self.ui_size_calc_combo.setEnabled(False)
self.ui_depth_slider.setEnabled(False)
self.ui_target_class_combo.setEnabled(False)
self.ui_zoom_slider.setEnabled(False)
yield
finally:
self.ui_size_calc_combo.setEnabled(True)
self.ui_depth_slider.setEnabled(True)
self.ui_target_class_combo.setEnabled(True)
self.ui_zoom_slider.setEnabled(True)
def _draw_trees(self):
self.grid_items, self.ptrees = [], []
num_trees = len(self.forest_adapter.get_trees())
with self.progressBar(num_trees) as prg, self.disable_ui():
for tree in self.forest_adapter.get_trees():
ptree = PythagorasTreeViewer(
None,
tree,
interactive=False,
padding=100,
target_class_index=self.target_class_index,
weight_adjustment=self.SIZE_CALCULATION[
self.size_calc_idx][1])
grid_item = GridItem(ptree,
self.grid,
max_size=self._calculate_zoom(self.zoom))
# We don't want to show flickering while the trees are being
grid_item.setVisible(False)
self.grid_items.append(grid_item)
self.ptrees.append(ptree)
prg.advance()
self.grid.set_items(self.grid_items)
# This is necessary when adding items for the first time
if self.grid:
width = (self.view.width() -
self.view.verticalScrollBar().width())
self.grid.reflow(width)
self.grid.setPreferredWidth(width)
# After drawing is complete, we show the trees
for grid_item in self.grid_items:
grid_item.setVisible(True)
@staticmethod
def _calculate_zoom(zoom_level):
"""Calculate the max size for grid items from zoom level setting."""
return zoom_level * 5
def onDeleteWidget(self):
"""When deleting the widget."""
super().onDeleteWidget()
self.clear()
def commit(self):
"""Commit the selected tree to output."""
if self.__prevent_commit:
return
if not self.scene.selectedItems():
self.send('Tree', None)
# The selected tree index should only reset when model changes
if self.model is None:
self.selected_tree_index = -1
return
selected_item = self.scene.selectedItems()[0]
self.selected_tree_index = self.grid_items.index(selected_item)
tree = self.model.trees[self.selected_tree_index]
tree.instances = self.instances
tree.meta_target_class_index = self.target_class_index
tree.meta_size_calc_idx = self.size_calc_idx
tree.meta_depth_limit = self.depth_limit
self.send('Tree', tree)
def send_report(self):
"""Send report."""
self.report_plot()
def _update_scene_rect(self):
self.scene.setSceneRect(self.scene.itemsBoundingRect())
def _update_target_class_combo(self):
self._clear_target_class_combo()
label = [
x for x in self.ui_target_class_combo.parent().children()
if isinstance(x, QLabel)
][0]
if self.instances.domain.has_discrete_class:
label_text = 'Target class'
values = [
c.title() for c in self.instances.domain.class_vars[0].values
]
values.insert(0, 'None')
else:
label_text = 'Node color'
values = list(ContinuousTreeNode.COLOR_METHODS.keys())
label.setText(label_text)
self.ui_target_class_combo.addItems(values)
self.ui_target_class_combo.setCurrentIndex(self.target_class_index)
def resizeEvent(self, ev):
width = (self.view.width() - self.view.verticalScrollBar().width())
self.grid.reflow(width)
self.grid.setPreferredWidth(width)
super().resizeEvent(ev)
return QRectF(0, 0, self.width, self.height)
def sizeHint(self, which, constraint):
return QSizeF(self.width, self.height)
def sizePolicy(self):
return QSizePolicy(QSizePolicy.Expanding, QSizePolicy.Expanding)
if __name__ == '__main__':
import sys
from Orange.data.table import Table
from AnyQt.QtWidgets import ( # pylint: disable=ungrouped-imports
QGraphicsView, QGraphicsScene, QApplication, QWidget
)
app = QApplication(sys.argv)
widget = QWidget()
widget.resize(500, 300)
scene = QGraphicsScene(widget)
view = QGraphicsView(scene, widget)
dataset = Table(sys.argv[1] if len(sys.argv) > 1 else 'iris')
histogram = Histogram(
dataset, variable=0, height=300, width=500, n_bins=20, bar_spacing=2,
border=(0, 0, 5, 0), border_color='#000', color_attribute='iris',
)
scene.addItem(histogram)
widget.show()
app.exec_()
