def create_model(rows, columns):
model = QStandardItemModel()
model.setRowCount(rows)
model.setColumnCount(columns)
for i in range(rows):
for j in range(columns):
model.setItemData(model.index(i, j), {
Qt.DisplayRole: f"{i}x{j}",
Qt.UserRole: i * j,
})
return model
def test_blockselectionmodel(self):
model = QStandardItemModel()
model.setRowCount(4)
model.setColumnCount(4)
sel = BlockSelectionModel(model)
sel.select(model.index(0, 0), BlockSelectionModel.Select)
self.assertSetEqual(selected(sel), {(0, 0)})
sel.select(model.index(0, 1), BlockSelectionModel.Select)
self.assertSetEqual(selected(sel), {(0, 0), (0, 1)})
sel.select(model.index(1, 1), BlockSelectionModel.Select)
self.assertSetEqual(selected(sel), {(0, 0), (0, 1), (1, 0), (1, 1)})
sel.select(model.index(0, 0), BlockSelectionModel.Deselect)
self.assertSetEqual(selected(sel), {(1, 1)})
sel.select(model.index(3, 3), BlockSelectionModel.ClearAndSelect)
self.assertSetEqual(selected(sel), {(3, 3)})
def test_symmetricselectionmodel(self):
model = QStandardItemModel()
model.setRowCount(4)
model.setColumnCount(4)
sel = SymmetricSelectionModel(model)
sel.select(model.index(0, 0), BlockSelectionModel.Select)
self.assertSetEqual(selected(sel), {(0, 0)})
sel.select(model.index(0, 2), BlockSelectionModel.Select)
self.assertSetEqual(selected(sel), {(0, 0), (0, 2), (2, 0), (2, 2)})
sel.select(model.index(0, 0), BlockSelectionModel.Deselect)
self.assertSetEqual(selected(sel), {(2, 2)})
sel.select(model.index(2, 3), BlockSelectionModel.ClearAndSelect)
self.assertSetEqual(selected(sel), {(2, 2), (2, 3), (3, 2), (3, 3)})
self.assertSetEqual(set(sel.selectedItems()), {2, 3})
sel.setSelectedItems([1, 2])
self.assertSetEqual(set(sel.selectedItems()), {1, 2})
def test_table_view_selection_finished(self):
model = QStandardItemModel()
model.setRowCount(10)
model.setColumnCount(4)
view = TableView()
view.setModel(model)
view.adjustSize()
spy = QSignalSpy(view.selectionFinished)
rect0 = view.visualRect(model.index(0, 0))
rect4 = view.visualRect(model.index(4, 2))
QTest.mousePress(
view.viewport(), Qt.LeftButton, Qt.NoModifier, rect0.center(),
)
self.assertEqual(len(spy), 0)
QTest.mouseRelease(
view.viewport(), Qt.LeftButton, Qt.NoModifier, rect4.center(),
)
self.assertEqual(len(spy), 1)
def test_header_view_clickable(self):
model = QStandardItemModel()
model.setColumnCount(3)
header = HeaderView(Qt.Horizontal)
header.setModel(model)
header.setSectionsClickable(True)
header.adjustSize()
pos = header.sectionViewportPosition(0)
size = header.sectionSize(0)
# center of first section
point = QPoint(pos + size // 2, header.viewport().height() / 2)
QTest.mousePress(header.viewport(), Qt.LeftButton, Qt.NoModifier, point)
opt = QStyleOptionHeader()
header.initStyleOptionForIndex(opt, 0)
self.assertTrue(opt.state & QStyle.State_Sunken)
QTest.mouseRelease(header.viewport(), Qt.LeftButton, Qt.NoModifier, point)
opt = QStyleOptionHeader()
header.initStyleOptionForIndex(opt, 0)
self.assertFalse(opt.state & QStyle.State_Sunken)
class OWConfusionMatrix(widget.OWWidget):
"""Confusion matrix widget"""
name = "Confusion Matrix"
description = "Display a confusion matrix constructed from " \
"the results of classifier evaluations."
icon = "icons/ConfusionMatrix.svg"
priority = 1001
class Inputs:
evaluation_results = Input("Evaluation Results",
Orange.evaluation.Results)
class Outputs:
selected_data = Output("Selected Data",
Orange.data.Table,
default=True)
annotated_data = Output(ANNOTATED_DATA_SIGNAL_NAME, Orange.data.Table)
quantities = [
"Number of instances", "Proportion of predicted",
"Proportion of actual"
]
settings_version = 1
settingsHandler = settings.ClassValuesContextHandler()
selected_learner = settings.Setting([0], schema_only=True)
selection = settings.ContextSetting(set())
selected_quantity = settings.Setting(0)
append_predictions = settings.Setting(True)
append_probabilities = settings.Setting(False)
autocommit = settings.Setting(True)
UserAdviceMessages = [
widget.Message(
"Clicking on cells or in headers outputs the corresponding "
"data instances", "click_cell")
]
class Error(widget.OWWidget.Error):
no_regression = Msg("Confusion Matrix cannot show regression results.")
invalid_values = Msg(
"Evaluation Results input contains invalid values")
def __init__(self):
super().__init__()
self.data = None
self.results = None
self.learners = []
self.headers = []
self.learners_box = gui.listBox(self.controlArea,
self,
"selected_learner",
"learners",
box=True,
callback=self._learner_changed)
self.outputbox = gui.vBox(self.controlArea, "Output")
box = gui.hBox(self.outputbox)
gui.checkBox(box,
self,
"append_predictions",
"Predictions",
callback=self._invalidate)
gui.checkBox(box,
self,
"append_probabilities",
"Probabilities",
callback=self._invalidate)
gui.auto_commit(self.outputbox,
self,
"autocommit",
"Send Selected",
"Send Automatically",
box=False)
self.mainArea.layout().setContentsMargins(0, 0, 0, 0)
box = gui.vBox(self.mainArea, box=True)
sbox = gui.hBox(box)
gui.rubber(sbox)
gui.comboBox(sbox,
self,
"selected_quantity",
items=self.quantities,
label="Show: ",
orientation=Qt.Horizontal,
callback=self._update)
self.tablemodel = QStandardItemModel(self)
view = self.tableview = QTableView(
editTriggers=QTableView.NoEditTriggers)
view.setModel(self.tablemodel)
view.horizontalHeader().hide()
view.verticalHeader().hide()
view.horizontalHeader().setMinimumSectionSize(60)
view.selectionModel().selectionChanged.connect(self._invalidate)
view.setShowGrid(False)
view.setItemDelegate(BorderedItemDelegate(Qt.white))
view.setSizePolicy(QSizePolicy.MinimumExpanding,
QSizePolicy.MinimumExpanding)
view.clicked.connect(self.cell_clicked)
box.layout().addWidget(view)
selbox = gui.hBox(box)
gui.button(selbox,
self,
"Select Correct",
callback=self.select_correct,
autoDefault=False)
gui.button(selbox,
self,
"Select Misclassified",
callback=self.select_wrong,
autoDefault=False)
gui.button(selbox,
self,
"Clear Selection",
callback=self.select_none,
autoDefault=False)
def sizeHint(self):
"""Initial size"""
return QSize(750, 340)
def _item(self, i, j):
return self.tablemodel.item(i, j) or QStandardItem()
def _set_item(self, i, j, item):
self.tablemodel.setItem(i, j, item)
def _init_table(self, nclasses):
item = self._item(0, 2)
item.setData("Predicted", Qt.DisplayRole)
item.setTextAlignment(Qt.AlignCenter)
item.setFlags(Qt.NoItemFlags)
self._set_item(0, 2, item)
item = self._item(2, 0)
item.setData("Actual", Qt.DisplayRole)
item.setTextAlignment(Qt.AlignHCenter | Qt.AlignBottom)
item.setFlags(Qt.NoItemFlags)
self.tableview.setItemDelegateForColumn(0, gui.VerticalItemDelegate())
self._set_item(2, 0, item)
self.tableview.setSpan(0, 2, 1, nclasses)
self.tableview.setSpan(2, 0, nclasses, 1)
font = self.tablemodel.invisibleRootItem().font()
bold_font = QFont(font)
bold_font.setBold(True)
for i in (0, 1):
for j in (0, 1):
item = self._item(i, j)
item.setFlags(Qt.NoItemFlags)
self._set_item(i, j, item)
for p, label in enumerate(self.headers):
for i, j in ((1, p + 2), (p + 2, 1)):
item = self._item(i, j)
item.setData(label, Qt.DisplayRole)
item.setFont(bold_font)
item.setTextAlignment(Qt.AlignRight | Qt.AlignVCenter)
item.setFlags(Qt.ItemIsEnabled)
if p < len(self.headers) - 1:
item.setData("br"[j == 1], BorderRole)
item.setData(QColor(192, 192, 192), BorderColorRole)
self._set_item(i, j, item)
hor_header = self.tableview.horizontalHeader()
if len(' '.join(self.headers)) < 120:
hor_header.setSectionResizeMode(QHeaderView.ResizeToContents)
else:
hor_header.setDefaultSectionSize(60)
self.tablemodel.setRowCount(nclasses + 3)
self.tablemodel.setColumnCount(nclasses + 3)
@Inputs.evaluation_results
def set_results(self, results):
"""Set the input results."""
prev_sel_learner = self.selected_learner.copy()
self.clear()
self.warning()
self.closeContext()
data = None
if results is not None and results.data is not None:
data = results.data[results.row_indices]
if data is not None and not data.domain.has_discrete_class:
self.Error.no_regression()
data = results = None
else:
self.Error.no_regression.clear()
nan_values = False
if results is not None:
assert isinstance(results, Orange.evaluation.Results)
if np.any(np.isnan(results.actual)) or \
np.any(np.isnan(results.predicted)):
# Error out here (could filter them out with a warning
# instead).
nan_values = True
results = data = None
if nan_values:
self.Error.invalid_values()
else:
self.Error.invalid_values.clear()
self.results = results
self.data = data
if data is not None:
class_values = data.domain.class_var.values
elif results is not None:
raise NotImplementedError
if results is None:
self.report_button.setDisabled(True)
else:
self.report_button.setDisabled(False)
nmodels = results.predicted.shape[0]
self.headers = class_values + \
[unicodedata.lookup("N-ARY SUMMATION")]
# NOTE: The 'learner_names' is set in 'Test Learners' widget.
if hasattr(results, "learner_names"):
self.learners = results.learner_names
else:
self.learners = [
"Learner #{}".format(i + 1) for i in range(nmodels)
]
self._init_table(len(class_values))
self.openContext(data.domain.class_var)
if not prev_sel_learner or prev_sel_learner[0] >= len(
self.learners):
if self.learners:
self.selected_learner[:] = [0]
else:
self.selected_learner[:] = prev_sel_learner
self._update()
self._set_selection()
self.unconditional_commit()
def clear(self):
"""Reset the widget, clear controls"""
self.results = None
self.data = None
self.tablemodel.clear()
self.headers = []
# Clear learners last. This action will invoke `_learner_changed`
self.learners = []
def select_correct(self):
"""Select the diagonal elements of the matrix"""
selection = QItemSelection()
n = self.tablemodel.rowCount()
for i in range(2, n):
index = self.tablemodel.index(i, i)
selection.select(index, index)
self.tableview.selectionModel().select(
selection, QItemSelectionModel.ClearAndSelect)
def select_wrong(self):
"""Select the off-diagonal elements of the matrix"""
selection = QItemSelection()
n = self.tablemodel.rowCount()
for i in range(2, n):
for j in range(i + 1, n):
index = self.tablemodel.index(i, j)
selection.select(index, index)
index = self.tablemodel.index(j, i)
selection.select(index, index)
self.tableview.selectionModel().select(
selection, QItemSelectionModel.ClearAndSelect)
def select_none(self):
"""Reset selection"""
self.tableview.selectionModel().clear()
def cell_clicked(self, model_index):
"""Handle cell click event"""
i, j = model_index.row(), model_index.column()
if not i or not j:
return
n = self.tablemodel.rowCount()
index = self.tablemodel.index
selection = None
if i == j == 1 or i == j == n - 1:
selection = QItemSelection(index(2, 2), index(n - 1, n - 1))
elif i in (1, n - 1):
selection = QItemSelection(index(2, j), index(n - 1, j))
elif j in (1, n - 1):
selection = QItemSelection(index(i, 2), index(i, n - 1))
if selection is not None:
self.tableview.selectionModel().select(
selection, QItemSelectionModel.ClearAndSelect)
def _prepare_data(self):
indices = self.tableview.selectedIndexes()
indices = {(ind.row() - 2, ind.column() - 2) for ind in indices}
actual = self.results.actual
learner_name = self.learners[self.selected_learner[0]]
predicted = self.results.predicted[self.selected_learner[0]]
selected = [
i for i, t in enumerate(zip(actual, predicted)) if t in indices
]
extra = []
class_var = self.data.domain.class_var
metas = self.data.domain.metas
if self.append_predictions:
extra.append(predicted.reshape(-1, 1))
var = Orange.data.DiscreteVariable(
"{}({})".format(class_var.name, learner_name),
class_var.values)
metas = metas + (var, )
if self.append_probabilities and \
self.results.probabilities is not None:
probs = self.results.probabilities[self.selected_learner[0]]
extra.append(np.array(probs, dtype=object))
pvars = [
Orange.data.ContinuousVariable("p({})".format(value))
for value in class_var.values
]
metas = metas + tuple(pvars)
domain = Orange.data.Domain(self.data.domain.attributes,
self.data.domain.class_vars, metas)
data = self.data.transform(domain)
if len(extra):
data.metas[:, len(self.data.domain.metas):] = \
np.hstack(tuple(extra))
data.name = learner_name
if selected:
annotated_data = create_annotated_table(data, selected)
data = data[selected]
else:
annotated_data = create_annotated_table(data, [])
data = None
return data, annotated_data
def commit(self):
"""Output data instances corresponding to selected cells"""
if self.results is not None and self.data is not None \
and self.selected_learner:
data, annotated_data = self._prepare_data()
else:
data = None
annotated_data = None
self.Outputs.selected_data.send(data)
self.Outputs.annotated_data.send(annotated_data)
def _invalidate(self):
indices = self.tableview.selectedIndexes()
self.selection = {(ind.row() - 2, ind.column() - 2) for ind in indices}
self.commit()
def _set_selection(self):
selection = QItemSelection()
index = self.tableview.model().index
for row, col in self.selection:
sel = index(row + 2, col + 2)
selection.select(sel, sel)
self.tableview.selectionModel().select(
selection, QItemSelectionModel.ClearAndSelect)
def _learner_changed(self):
self._update()
self._set_selection()
self.commit()
def _update(self):
def _isinvalid(x):
return isnan(x) or isinf(x)
# Update the displayed confusion matrix
if self.results is not None and self.selected_learner:
cmatrix = confusion_matrix(self.results, self.selected_learner[0])
colsum = cmatrix.sum(axis=0)
rowsum = cmatrix.sum(axis=1)
n = len(cmatrix)
diag = np.diag_indices(n)
colors = cmatrix.astype(np.double)
colors[diag] = 0
if self.selected_quantity == 0:
normalized = cmatrix.astype(np.int)
formatstr = "{}"
div = np.array([colors.max()])
else:
if self.selected_quantity == 1:
normalized = 100 * cmatrix / colsum
div = colors.max(axis=0)
else:
normalized = 100 * cmatrix / rowsum[:, np.newaxis]
div = colors.max(axis=1)[:, np.newaxis]
formatstr = "{:2.1f} %"
div[div == 0] = 1
colors /= div
colors[diag] = normalized[diag] / normalized[diag].max()
for i in range(n):
for j in range(n):
val = normalized[i, j]
col_val = colors[i, j]
item = self._item(i + 2, j + 2)
item.setData(
"NA" if _isinvalid(val) else formatstr.format(val),
Qt.DisplayRole)
bkcolor = QColor.fromHsl(
[0, 240][i == j], 160,
255 if _isinvalid(col_val) else int(255 -
30 * col_val))
item.setData(QBrush(bkcolor), Qt.BackgroundRole)
item.setData("trbl", BorderRole)
item.setToolTip("actual: {}\npredicted: {}".format(
self.headers[i], self.headers[j]))
item.setTextAlignment(Qt.AlignRight | Qt.AlignVCenter)
item.setFlags(Qt.ItemIsEnabled | Qt.ItemIsSelectable)
self._set_item(i + 2, j + 2, item)
bold_font = self.tablemodel.invisibleRootItem().font()
bold_font.setBold(True)
def _sum_item(value, border=""):
item = QStandardItem()
item.setData(value, Qt.DisplayRole)
item.setTextAlignment(Qt.AlignRight | Qt.AlignVCenter)
item.setFlags(Qt.ItemIsEnabled)
item.setFont(bold_font)
item.setData(border, BorderRole)
item.setData(QColor(192, 192, 192), BorderColorRole)
return item
for i in range(n):
self._set_item(n + 2, i + 2, _sum_item(int(colsum[i]), "t"))
self._set_item(i + 2, n + 2, _sum_item(int(rowsum[i]), "l"))
self._set_item(n + 2, n + 2, _sum_item(int(rowsum.sum())))
def send_report(self):
"""Send report"""
if self.results is not None and self.selected_learner:
self.report_table(
"Confusion matrix for {} (showing {})".format(
self.learners[self.selected_learner[0]],
self.quantities[self.selected_quantity].lower()),
self.tableview)
@classmethod
def migrate_settings(cls, settings, version):
if not version:
# For some period of time the 'selected_learner' property was
# changed from List[int] -> int
# (commit 4e49bb3fd0e11262f3ebf4b1116a91a4b49cc982) and then back
# again (commit 8a492d79a2e17154a0881e24a05843406c8892c0)
if "selected_learner" in settings and \
isinstance(settings["selected_learner"], int):
settings["selected_learner"] = [settings["selected_learner"]]
class ScoreTable(OWComponent, QObject):
shown_scores = \
Setting(set(chain(*BUILTIN_SCORERS_ORDER.values())))
shownScoresChanged = Signal()
class ItemDelegate(QStyledItemDelegate):
def sizeHint(self, *args):
size = super().sizeHint(*args)
return QSize(size.width(), size.height() + 6)
def displayText(self, value, locale):
if isinstance(value, float):
return f"{value:.3f}"
else:
return super().displayText(value, locale)
def __init__(self, master):
QObject.__init__(self)
OWComponent.__init__(self, master)
self.view = gui.TableView(wordWrap=True,
editTriggers=gui.TableView.NoEditTriggers)
header = self.view.horizontalHeader()
header.setSectionResizeMode(QHeaderView.ResizeToContents)
header.setDefaultAlignment(Qt.AlignCenter)
header.setStretchLastSection(False)
header.setContextMenuPolicy(Qt.CustomContextMenu)
header.customContextMenuRequested.connect(self.show_column_chooser)
self.model = QStandardItemModel(master)
self.model.setHorizontalHeaderLabels(["Method"])
self.sorted_model = ScoreModel()
self.sorted_model.setSourceModel(self.model)
self.view.setModel(self.sorted_model)
self.view.setItemDelegate(self.ItemDelegate())
def _column_names(self):
return (self.model.horizontalHeaderItem(section).data(Qt.DisplayRole)
for section in range(1, self.model.columnCount()))
def show_column_chooser(self, pos):
# pylint doesn't know that self.shown_scores is a set, not a Setting
# pylint: disable=unsupported-membership-test
def update(col_name, checked):
if checked:
self.shown_scores.add(col_name)
else:
self.shown_scores.remove(col_name)
self._update_shown_columns()
menu = QMenu()
header = self.view.horizontalHeader()
for col_name in self._column_names():
action = menu.addAction(col_name)
action.setCheckable(True)
action.setChecked(col_name in self.shown_scores)
action.triggered.connect(partial(update, col_name))
menu.exec(header.mapToGlobal(pos))
def _update_shown_columns(self):
# pylint doesn't know that self.shown_scores is a set, not a Setting
# pylint: disable=unsupported-membership-test
header = self.view.horizontalHeader()
for section, col_name in enumerate(self._column_names(), start=1):
header.setSectionHidden(section, col_name not in self.shown_scores)
self.view.resizeColumnsToContents()
self.shownScoresChanged.emit()
def update_header(self, scorers):
# Set the correct horizontal header labels on the results_model.
self.model.setColumnCount(3 + len(scorers))
self.model.setHorizontalHeaderItem(0, QStandardItem("Model"))
self.model.setHorizontalHeaderItem(1, QStandardItem("Train time [s]"))
self.model.setHorizontalHeaderItem(2, QStandardItem("Test time [s]"))
for col, score in enumerate(scorers, start=3):
item = QStandardItem(score.name)
item.setToolTip(score.long_name)
self.model.setHorizontalHeaderItem(col, item)
self._update_shown_columns()
class ContingencyTable(QTableView):
"""
A contingency table widget which can be used wherever ``QTableView`` could be used.
Parameters
----------
parent : Orange.widgets.widget.OWWidget
The containing widget to which the table is connected.
Attributes
----------
classesv : :obj:`list` of :obj:`str`
Vertical class headers.
classesh : :obj:`list` of :obj:`str`
Horizontal class headers.
headerv : :obj:`str`, optional
Vertical top header.
headerh : :obj:`str`, optional
Horizontal top header.
corner_string : str
String that is top right and bottom left corner of the table.
Default is ``unicodedata.lookup("N-ARY SUMMATION")``.
"""
def __init__(self, parent):
super().__init__(editTriggers=QTableView.NoEditTriggers)
self.bold_headers = None
self.circles = False
self.classesv = None
self.classesh = None
self.headerv = None
self.headerh = None
self.parent = parent
self.corner_string = unicodedata.lookup("N-ARY SUMMATION")
self.tablemodel = QStandardItemModel(self)
self.setModel(self.tablemodel)
self.horizontalHeader().hide()
self.verticalHeader().hide()
self.horizontalHeader().setMinimumSectionSize(60)
self.setShowGrid(False)
self.setSizePolicy(QSizePolicy.MinimumExpanding,
QSizePolicy.MinimumExpanding)
self.clicked.connect(self._cell_clicked)
def mouseReleaseEvent(self, e):
super().mouseReleaseEvent(e)
self.parent._invalidate()
def keyPressEvent(self, event):
super().keyPressEvent(event)
self.parent._invalidate()
def _cell_clicked(self, model_index):
"""Handle cell click event"""
i, j = model_index.row(), model_index.column()
if not i or not j:
return
n = self.tablemodel.rowCount()
m = self.tablemodel.columnCount()
index = self.tablemodel.index
selection = None
if i == j == 1 or not self.circles and i == n - 1 and j == m - 1:
selection = QItemSelection(index(2, 2), index(n - 1, m - 1))
elif i == 1 or not self.circles and i == n - 1:
selection = QItemSelection(index(2, j), index(n - 1, j))
elif j == 1 or not self.circles and j == m - 1:
selection = QItemSelection(index(i, 2), index(i, m - 1))
if selection is not None:
self.selectionModel().select(selection,
QItemSelectionModel.ClearAndSelect)
def _item(self, i, j):
return self.tablemodel.item(i, j) or QStandardItem()
def _set_item(self, i, j, item):
self.tablemodel.setItem(i, j, item)
def set_variables(self, variablev, variableh, **kwargs):
"""
Sets class headers and top headers and initializes table structure.
Parameters
----------
variablev : Orange.data.variable.DiscreteVariable
Class headers are set to ``variablev.values``, top header is set to ``variablev.name``.
variableh : Orange.data.variable.DiscreteVariable
Class headers are set to ``variableh.values``, top header is set to ``variableh.name``.
"""
self.classesv = variablev.values
self.classesh = variableh.values
self.headerv = variablev.name
self.headerh = variableh.name
self.initialize(**kwargs)
def set_headers(self,
classesv,
classesh,
headerv=None,
headerh=None,
**kwargs):
"""
Sets class headers and top headers and initializes table structure.
Parameters
----------
classesv : :obj:`list` of :obj:`str`
Vertical class headers.
classesh : :obj:`list` of :obj:`str`
Horizontal class headers.
headerv : :obj:`str`, optional
Vertical top header.
headerh : :obj:`str`, optional
Horizontal top header.
"""
self.classesv = classesv
self.classesh = classesh
self.headerv = headerv
self.headerh = headerh
self.initialize(**kwargs)
def _style_cells(self):
"""
Style all cells.
"""
if self.circles:
self.setItemDelegate(CircleItemDelegate(Qt.white))
else:
self.setItemDelegate(BorderedItemDelegate(Qt.white))
item = self._item(0, 2)
item.setData(self.headerh, Qt.DisplayRole)
item.setTextAlignment(Qt.AlignCenter)
item.setFlags(Qt.NoItemFlags)
self._set_item(0, 2, item)
item = self._item(2, 0)
item.setData(self.headerv, Qt.DisplayRole)
item.setTextAlignment(Qt.AlignHCenter | Qt.AlignBottom)
item.setFlags(Qt.NoItemFlags)
self.setItemDelegateForColumn(0, gui.VerticalItemDelegate())
self._set_item(2, 0, item)
self.setSpan(0, 2, 1, len(self.classesh) + 1)
self.setSpan(2, 0, len(self.classesv) + 1, 1)
for i in (0, 1):
for j in (0, 1):
item = self._item(i, j)
item.setFlags(Qt.NoItemFlags)
self._set_item(i, j, item)
def _initialize_headers(self):
"""
Fill headers with content and style them.
"""
font = self.tablemodel.invisibleRootItem().font()
bold_font = QFont(font)
bold_font.setBold(True)
for headers, ix in ((self.classesv + [self.corner_string], lambda p:
(p + 2, 1)),
(self.classesh + [self.corner_string], lambda p:
(1, p + 2))):
for p, label in enumerate(headers):
i, j = ix(p)
item = self._item(i, j)
item.setData(label, Qt.DisplayRole)
if self.bold_headers:
item.setFont(bold_font)
if not (i == 1 and self.circles):
item.setTextAlignment(Qt.AlignRight | Qt.AlignVCenter)
item.setFlags(Qt.ItemIsEnabled)
if p < len(headers) - 1:
item.setData("br"[j == 1], BorderRole)
item.setData(QColor(192, 192, 192), BorderColorRole)
else:
item.setData("", BorderRole)
self._set_item(i, j, item)
def _resize(self):
"""
Resize table to fit new contents and style.
"""
if self.circles:
self.resizeRowToContents(1)
self.horizontalHeader().setDefaultSectionSize(self.rowHeight(2))
self.resizeColumnToContents(1)
self.tablemodel.setRowCount(len(self.classesv) + 2)
self.tablemodel.setColumnCount(len(self.classesh) + 2)
else:
if len(' '.join(self.classesh + [self.corner_string])) < 120:
self.horizontalHeader().setSectionResizeMode(
QHeaderView.ResizeToContents)
else:
self.horizontalHeader().setDefaultSectionSize(60)
self.tablemodel.setRowCount(len(self.classesv) + 3)
self.tablemodel.setColumnCount(len(self.classesh) + 3)
def initialize(self, circles=False, bold_headers=True):
"""
Initializes table structure. Class headers must be set beforehand.
Parameters
----------
circles : :obj:`bool`, optional
Turns on circle display. All table values should be between 0 and 1 (inclusive). Defaults to False.
bold_headers : :obj:`bool`, optional
Whether the headers are bold or not. Defaults to True.
"""
assert self.classesv is not None and self.classesh is not None
self.circles = circles
self.bold_headers = bold_headers
self._style_cells()
self._initialize_headers()
self._resize()
def get_selection(self):
"""
Get indexes of selected cells.
Returns
-------
:obj:`set` of :obj:`tuple` of :obj:`int`
Set of pairs of indexes.
"""
return {(ind.row() - 2, ind.column() - 2)
for ind in self.selectedIndexes()}
def set_selection(self, indexes):
"""
Set indexes of selected cells.
Parameters
----------
indexes : :obj:`set` of :obj:`tuple` of :obj:`int`
Set of pairs of indexes.
"""
selection = QItemSelection()
index = self.model().index
for row, col in indexes:
sel = index(row + 2, col + 2)
selection.select(sel, sel)
self.selectionModel().select(selection,
QItemSelectionModel.ClearAndSelect)
def _set_sums(self, colsum, rowsum):
"""
Set content of cells on bottom and right edge.
Parameters
----------
colsum : numpy.array
Content of cells on bottom edge.
rowsum : numpy.array
Content of cells on right edge.
"""
bold_font = self.tablemodel.invisibleRootItem().font()
bold_font.setBold(True)
def _sum_item(value, border=""):
item = QStandardItem()
item.setData(value, Qt.DisplayRole)
item.setTextAlignment(Qt.AlignRight | Qt.AlignVCenter)
item.setFlags(Qt.ItemIsEnabled)
item.setFont(bold_font)
item.setData(border, BorderRole)
item.setData(QColor(192, 192, 192), BorderColorRole)
return item
for i in range(len(self.classesh)):
self._set_item(
len(self.classesv) + 2, i + 2, _sum_item(int(colsum[i]), "t"))
for i in range(len(self.classesv)):
self._set_item(i + 2,
len(self.classesh) + 2,
_sum_item(int(rowsum[i]), "l"))
self._set_item(
len(self.classesv) + 2,
len(self.classesh) + 2, _sum_item(int(rowsum.sum())))
def _set_values(self, matrix, colors, formatstr, tooltip):
"""
Set content of cells which aren't headers and don't represent aggregate values.
Parameters
----------
matrix : numpy.array
2D array to be set as data.
colors : :obj:`numpy.array`
2D array with color values.
formatstr : :obj:`str`, optional
Format string for cell data.
tooltip : :obj:`(int, int) -> str`
Function which takes vertical index and horizontal index as arguments and returns
desired tooltip as a string.
"""
def _isinvalid(x):
return isnan(x) or isinf(x)
for i in range(len(self.classesv)):
for j in range(len(self.classesh)):
val = matrix[i, j]
col_val = float('nan') if colors is None else colors[i, j]
item = QStandardItem()
if self.circles:
item.setData(val, CircleAreaRole)
else:
item.setData(
"NA" if _isinvalid(val) else formatstr.format(val),
Qt.DisplayRole)
bkcolor = QColor.fromHsl(
[0, 240][i == j], 160,
255 if _isinvalid(col_val) else int(255 -
30 * col_val))
item.setData(QBrush(bkcolor), Qt.BackgroundRole)
item.setData("trbl", BorderRole)
if tooltip is not None:
item.setToolTip(tooltip(i, j))
item.setTextAlignment(Qt.AlignRight | Qt.AlignVCenter)
item.setFlags(Qt.ItemIsEnabled | Qt.ItemIsSelectable)
self._set_item(i + 2, j + 2, item)
def update_table(self,
matrix,
colsum=None,
rowsum=None,
colors=None,
formatstr="{}",
tooltip=None):
"""
Sets ``matrix`` as data of the table.
Parameters
----------
matrix : numpy.array
2D array to be set as data.
colsum : :obj:`numpy.array`, optional
1D optional array with aggregate values of columns, defaults to sum.
rowsum : :obj:`numpy.array`, optional
1D optional array with aggregate values of rows, defaults to sum.
colors : :obj:`numpy.array`, optional
2D array with color values, defaults to no color.
formatstr : :obj:`str`, optional
Format string for cell data, defaults to ``"{}"``.
tooltip : :obj:`(int, int) -> str`, optional
Function which takes vertical index and horizontal index as arguments and returns
desired tooltip as a string. Defaults to no tooltips.
"""
selected_indexes = self.get_selection()
self._set_values(matrix, colors, formatstr, tooltip)
if not self.circles:
if colsum is None:
colsum = matrix.sum(axis=0)
if rowsum is None:
rowsum = matrix.sum(axis=1)
self._set_sums(colsum, rowsum)
self.set_selection(selected_indexes)
def clear(self):
"""
Clears the table.
"""
self.tablemodel.clear()
def test_header(self):
model = QStandardItemModel()
hheader = HeaderView(Qt.Horizontal)
vheader = HeaderView(Qt.Vertical)
hheader.setSortIndicatorShown(True)
# paint with no model.
vheader.grab()
hheader.grab()
hheader.setModel(model)
vheader.setModel(model)
hheader.adjustSize()
vheader.adjustSize()
# paint with an empty model
vheader.grab()
hheader.grab()
model.setRowCount(1)
model.setColumnCount(1)
icon = QIcon(StampIconEngine("A", Qt.red))
model.setHeaderData(0, Qt.Horizontal, icon, Qt.DecorationRole)
model.setHeaderData(0, Qt.Vertical, icon, Qt.DecorationRole)
model.setHeaderData(0, Qt.Horizontal, QColor(Qt.blue), Qt.ForegroundRole)
model.setHeaderData(0, Qt.Vertical, QColor(Qt.blue), Qt.ForegroundRole)
model.setHeaderData(0, Qt.Horizontal, QColor(Qt.white), Qt.BackgroundRole)
model.setHeaderData(0, Qt.Vertical, QColor(Qt.white), Qt.BackgroundRole)
# paint with single col/row model
vheader.grab()
hheader.grab()
model.setRowCount(3)
model.setColumnCount(3)
hheader.adjustSize()
vheader.adjustSize()
# paint with single col/row model
vheader.grab()
hheader.grab()
hheader.setSortIndicator(0, Qt.AscendingOrder)
vheader.setHighlightSections(True)
hheader.setHighlightSections(True)
vheader.grab()
hheader.grab()
vheader.setSectionsClickable(True)
hheader.setSectionsClickable(True)
vheader.grab()
hheader.grab()
vheader.setTextElideMode(Qt.ElideRight)
hheader.setTextElideMode(Qt.ElideRight)
selmodel = QItemSelectionModel(model, model)
vheader.setSelectionModel(selmodel)
hheader.setSelectionModel(selmodel)
selmodel.select(model.index(1, 1), QItemSelectionModel.Rows | QItemSelectionModel.Select)
selmodel.select(model.index(1, 1), QItemSelectionModel.Columns | QItemSelectionModel.Select)
vheader.grab()
vheader.grab()
class OWKMeans(widget.OWWidget):
name = "k-Means"
description = "k-Means clustering algorithm with silhouette-based " \
"quality estimation."
icon = "icons/KMeans.svg"
priority = 2100
inputs = [("Data", Table, "set_data")]
outputs = [("Annotated Data", Table, widget.Default), ("Centroids", Table)]
class Error(widget.OWWidget.Error):
failed = widget.Msg("Clustering failed\nError: {}")
INIT_KMEANS, INIT_RANDOM = range(2)
INIT_METHODS = "Initialize with KMeans++", "Random initialization"
SILHOUETTE, INTERCLUSTER, DISTANCES = range(3)
SCORING_METHODS = [
("Silhouette", lambda km: km.silhouette, False, True),
("Inter-cluster distance", lambda km: km.inter_cluster, True, False),
("Distance to centroids", lambda km: km.inertia, True, False)
]
OUTPUT_CLASS, OUTPUT_ATTRIBUTE, OUTPUT_META = range(3)
OUTPUT_METHODS = ("Class", "Feature", "Meta")
resizing_enabled = False
k = Setting(3)
k_from = Setting(2)
k_to = Setting(8)
optimize_k = Setting(False)
max_iterations = Setting(300)
n_init = Setting(10)
smart_init = Setting(INIT_KMEANS)
scoring = Setting(SILHOUETTE)
append_cluster_ids = Setting(True)
place_cluster_ids = Setting(OUTPUT_CLASS)
output_name = Setting("Cluster")
auto_run = Setting(True)
def __init__(self):
super().__init__()
self.data = None
self.km = None
self.optimization_runs = []
box = gui.vBox(self.controlArea, "Number of Clusters")
layout = QGridLayout()
self.n_clusters = bg = gui.radioButtonsInBox(box,
self,
"optimize_k", [],
orientation=layout,
callback=self.update)
layout.addWidget(
gui.appendRadioButton(bg, "Fixed:", addToLayout=False), 1, 1)
sb = gui.hBox(None, margin=0)
self.fixedSpinBox = gui.spin(sb,
self,
"k",
minv=2,
maxv=30,
controlWidth=60,
alignment=Qt.AlignRight,
callback=self.update_k)
gui.rubber(sb)
layout.addWidget(sb, 1, 2)
layout.addWidget(
gui.appendRadioButton(bg, "Optimized from", addToLayout=False), 2,
1)
ftobox = gui.hBox(None)
ftobox.layout().setContentsMargins(0, 0, 0, 0)
layout.addWidget(ftobox)
gui.spin(ftobox,
self,
"k_from",
minv=2,
maxv=29,
controlWidth=60,
alignment=Qt.AlignRight,
callback=self.update_from)
gui.widgetLabel(ftobox, "to")
self.fixedSpinBox = gui.spin(ftobox,
self,
"k_to",
minv=3,
maxv=30,
controlWidth=60,
alignment=Qt.AlignRight,
callback=self.update_to)
gui.rubber(ftobox)
layout.addWidget(gui.widgetLabel(None, "Scoring: "), 5, 1,
Qt.AlignRight)
layout.addWidget(
gui.comboBox(None,
self,
"scoring",
label="Scoring",
items=list(zip(*self.SCORING_METHODS))[0],
callback=self.update), 5, 2)
box = gui.vBox(self.controlArea, "Initialization")
gui.comboBox(box,
self,
"smart_init",
items=self.INIT_METHODS,
callback=self.update)
layout = QGridLayout()
box2 = gui.widgetBox(box, orientation=layout)
box2.setSizePolicy(QSizePolicy.Maximum, QSizePolicy.Maximum)
layout.addWidget(gui.widgetLabel(None, "Re-runs: "), 0, 0,
Qt.AlignLeft)
sb = gui.hBox(None, margin=0)
layout.addWidget(sb, 0, 1)
gui.lineEdit(sb,
self,
"n_init",
controlWidth=60,
valueType=int,
validator=QIntValidator(),
callback=self.update)
layout.addWidget(gui.widgetLabel(None, "Maximal iterations: "), 1, 0,
Qt.AlignLeft)
sb = gui.hBox(None, margin=0)
layout.addWidget(sb, 1, 1)
gui.lineEdit(sb,
self,
"max_iterations",
controlWidth=60,
valueType=int,
validator=QIntValidator(),
callback=self.update)
box = gui.vBox(self.controlArea, "Output")
gui.comboBox(box,
self,
"place_cluster_ids",
label="Append cluster ID as:",
orientation=Qt.Horizontal,
callback=self.send_data,
items=self.OUTPUT_METHODS)
gui.lineEdit(box,
self,
"output_name",
label="Name:",
orientation=Qt.Horizontal,
callback=self.send_data)
gui.separator(self.buttonsArea, 30)
self.apply_button = gui.auto_commit(self.buttonsArea,
self,
"auto_run",
"Apply",
box=None,
commit=self.commit)
gui.rubber(self.controlArea)
self.table_model = QStandardItemModel(self)
self.table_model.setHorizontalHeaderLabels(["k", "Score"])
self.table_model.setColumnCount(2)
self.table_box = gui.vBox(self.mainArea,
"Optimization Report",
addSpace=0)
table = self.table_view = QTableView(self.table_box)
table.setHorizontalScrollMode(QTableView.ScrollPerPixel)
table.setHorizontalScrollBarPolicy(Qt.ScrollBarAlwaysOff)
table.setSelectionMode(QTableView.SingleSelection)
table.setSelectionBehavior(QTableView.SelectRows)
table.verticalHeader().hide()
self.bar_delegate = gui.ColoredBarItemDelegate(self, color=Qt.cyan)
table.setItemDelegateForColumn(1, self.bar_delegate)
table.setModel(self.table_model)
table.selectionModel().selectionChanged.connect(
self.table_item_selected)
table.setColumnWidth(0, 40)
table.setColumnWidth(1, 120)
table.horizontalHeader().setStretchLastSection(True)
self.setSizePolicy(QSizePolicy.Preferred, QSizePolicy.Preferred)
self.mainArea.setSizePolicy(QSizePolicy.Maximum, QSizePolicy.Preferred)
self.table_box.setSizePolicy(QSizePolicy.Fixed,
QSizePolicy.MinimumExpanding)
self.table_view.setSizePolicy(QSizePolicy.Preferred,
QSizePolicy.MinimumExpanding)
self.table_box.layout().addWidget(self.table_view)
self.hide_show_opt_results()
def adjustSize(self):
self.ensurePolished()
s = self.sizeHint()
self.resize(s)
def hide_show_opt_results(self):
[self.mainArea.hide, self.mainArea.show][self.optimize_k]()
QTimer.singleShot(100, self.adjustSize)
def sizeHint(self):
s = self.controlArea.sizeHint()
if self.optimize_k and not self.mainArea.isHidden():
s.setWidth(s.width() + self.mainArea.sizeHint().width() +
4 * self.childrenRect().x())
return s
def update_k(self):
self.optimize_k = False
self.update()
def update_from(self):
self.k_to = max(self.k_from + 1, self.k_to)
self.optimize_k = True
self.update()
def update_to(self):
self.k_from = min(self.k_from, self.k_to - 1)
self.optimize_k = True
self.update()
def set_optimization(self):
self.updateOptimizationGui()
self.update()
def check_data_size(self, n, msg_group):
msg_group.add_message(
"not_enough_data",
"Too few ({}) unique data instances for {} clusters")
if n > len(self.data):
msg_group.not_enough_data(len(self.data), n)
return False
else:
msg_group.not_enough_data.clear()
return True
def run_optimization(self):
# Disabling is needed since this function is not reentrant
# Fast clicking on, say, "To: " causes multiple calls
try:
self.controlArea.setDisabled(True)
self.optimization_runs = []
error = ""
if not self.check_data_size(self.k_from, self.Error):
return
self.check_data_size(self.k_to, self.Warning)
k_to = min(self.k_to, len(self.data))
kmeans = KMeans(
init=['random', 'k-means++'][self.smart_init],
n_init=self.n_init,
max_iter=self.max_iterations,
compute_silhouette_score=self.scoring == self.SILHOUETTE)
with self.progressBar(k_to - self.k_from + 1) as progress:
for k in range(self.k_from, k_to + 1):
progress.advance()
kmeans.params["n_clusters"] = k
try:
self.optimization_runs.append((k, kmeans(self.data)))
except BaseException as exc:
error = str(exc)
self.optimization_runs.append((k, error))
if all(
isinstance(score, str)
for _, score in self.optimization_runs):
self.Error.failed(error) # Report just the last error
self.optimization_runs = []
finally:
self.controlArea.setDisabled(False)
self.show_results()
self.send_data()
def cluster(self):
if not self.check_data_size(self.k, self.Error):
return
try:
self.km = KMeans(n_clusters=self.k,
init=['random', 'k-means++'][self.smart_init],
n_init=self.n_init,
max_iter=self.max_iterations)(self.data)
except BaseException as exc:
self.Error.failed(str(exc))
self.km = None
self.send_data()
def run(self):
self.clear_messages()
if not self.data:
return
if self.optimize_k:
self.run_optimization()
else:
self.cluster()
def commit(self):
self.run()
def show_results(self):
_, scoring_method, minimize, normal = self.SCORING_METHODS[
self.scoring]
k_scores = [(k,
scoring_method(run) if not isinstance(run, str) else run)
for k, run in self.optimization_runs]
scores = [score for _, score in k_scores if not isinstance(score, str)]
min_score, max_score = min(scores, default=0), max(scores, default=1)
best_score = min_score if minimize else max_score
if normal:
min_score, max_score = 0, 1
nplaces = 3
else:
nplaces = min(5,
np.floor(abs(math.log(max(max_score, 1e-10)))) + 2)
score_span = (max_score - min_score) or 1
self.bar_delegate.scale = (min_score, max_score)
self.bar_delegate.float_fmt = "%%.%if" % int(nplaces)
model = self.table_model
model.setRowCount(len(k_scores))
no_selection = True
for i, (k, score) in enumerate(k_scores):
item0 = model.item(i, 0) or QStandardItem()
item0.setData(k, Qt.DisplayRole)
item0.setTextAlignment(Qt.AlignCenter)
model.setItem(i, 0, item0)
item = model.item(i, 1) or QStandardItem()
if not isinstance(score, str):
item.setData(score, Qt.DisplayRole)
item.setData(None, Qt.ToolTipRole)
bar_ratio = 0.95 * (score - min_score) / score_span
item.setData(bar_ratio, gui.BarRatioRole)
if no_selection and score == best_score:
self.table_view.selectRow(i)
no_selection = False
color = Qt.black
flags = Qt.ItemIsEnabled | Qt.ItemIsSelectable
else:
item.setData("clustering failed", Qt.DisplayRole)
item.setData(score, Qt.ToolTipRole)
item.setData(None, gui.BarRatioRole)
color = Qt.gray
flags = Qt.NoItemFlags
item0.setData(QBrush(color), Qt.ForegroundRole)
item0.setFlags(flags)
item.setData(QBrush(color), Qt.ForegroundRole)
item.setFlags(flags)
model.setItem(i, 1, item)
self.table_view.resizeRowsToContents()
self.table_view.show()
if minimize:
self.table_box.setTitle("Scoring (smaller is better)")
else:
self.table_box.setTitle("Scoring (bigger is better)")
QTimer.singleShot(0, self.adjustSize)
def update(self):
self.hide_show_opt_results()
self.run()
def selected_row(self):
indices = self.table_view.selectedIndexes()
rows = {ind.row() for ind in indices}
if len(rows) == 1:
return rows.pop()
def table_item_selected(self):
row = self.selected_row()
if row is not None:
self.send_data()
def send_data(self):
if self.optimize_k:
row = self.selected_row() if self.optimization_runs else None
km = self.optimization_runs[row][1] if row is not None else None
else:
km = self.km
if not self.data or not km:
self.send("Annotated Data", None)
self.send("Centroids", None)
return
clust_var = DiscreteVariable(
self.output_name, values=["C%d" % (x + 1) for x in range(km.k)])
clust_ids = km(self.data)
domain = self.data.domain
attributes, classes = domain.attributes, domain.class_vars
meta_attrs = domain.metas
if self.place_cluster_ids == self.OUTPUT_CLASS:
if classes:
meta_attrs += classes
classes = [clust_var]
elif self.place_cluster_ids == self.OUTPUT_ATTRIBUTE:
attributes += (clust_var, )
else:
meta_attrs += (clust_var, )
domain = Domain(attributes, classes, meta_attrs)
new_table = Table.from_table(domain, self.data)
new_table.get_column_view(clust_var)[0][:] = clust_ids.X.ravel()
centroids = Table(Domain(km.pre_domain.attributes), km.centroids)
self.send("Annotated Data", new_table)
self.send("Centroids", centroids)
@check_sql_input
def set_data(self, data):
self.data = data
if data is None:
self.Error.clear()
self.Warning.clear()
self.table_model.setRowCount(0)
self.send("Annotated Data", None)
self.send("Centroids", None)
else:
self.data = data
self.run()
def send_report(self):
k_clusters = self.k
if self.optimize_k and self.optimization_runs and self.selected_row(
) is not None:
k_clusters = self.optimization_runs[self.selected_row()][1].k
self.report_items(
(("Number of clusters",
k_clusters), ("Optimization", self.optimize_k != 0
and "{}, {} re-runs limited to {} steps".format(
self.INIT_METHODS[self.smart_init].lower(),
self.n_init, self.max_iterations)),
("Cluster ID in output", self.append_cluster_ids
and "'{}' (as {})".format(
self.output_name,
self.OUTPUT_METHODS[self.place_cluster_ids].lower()))))
if self.data:
self.report_data("Data", self.data)
if self.optimize_k:
self.report_table(
"Scoring by {}".format(
self.SCORING_METHODS[self.scoring][0]),
self.table_view)
class OWConfusionMatrix(widget.OWWidget):
"""Confusion matrix widget"""
name = "Confusion Matrix"
description = "Display a confusion matrix constructed from " \
"the results of classifier evaluations."
icon = "icons/ConfusionMatrix.svg"
priority = 1001
class Inputs:
evaluation_results = Input("Evaluation Results", Orange.evaluation.Results)
class Outputs:
selected_data = Output("Selected Data", Orange.data.Table, default=True)
annotated_data = Output(ANNOTATED_DATA_SIGNAL_NAME, Orange.data.Table)
quantities = ["Number of instances",
"Proportion of predicted",
"Proportion of actual"]
settings_version = 1
settingsHandler = settings.ClassValuesContextHandler()
selected_learner = settings.Setting([0], schema_only=True)
selection = settings.ContextSetting(set())
selected_quantity = settings.Setting(0)
append_predictions = settings.Setting(True)
append_probabilities = settings.Setting(False)
autocommit = settings.Setting(True)
UserAdviceMessages = [
widget.Message(
"Clicking on cells or in headers outputs the corresponding "
"data instances",
"click_cell")]
class Error(widget.OWWidget.Error):
no_regression = Msg("Confusion Matrix cannot show regression results.")
invalid_values = Msg("Evaluation Results input contains invalid values")
def __init__(self):
super().__init__()
self.data = None
self.results = None
self.learners = []
self.headers = []
self.learners_box = gui.listBox(
self.controlArea, self, "selected_learner", "learners", box=True,
callback=self._learner_changed
)
self.outputbox = gui.vBox(self.controlArea, "Output")
box = gui.hBox(self.outputbox)
gui.checkBox(box, self, "append_predictions",
"Predictions", callback=self._invalidate)
gui.checkBox(box, self, "append_probabilities",
"Probabilities",
callback=self._invalidate)
gui.auto_commit(self.outputbox, self, "autocommit",
"Send Selected", "Send Automatically", box=False)
self.mainArea.layout().setContentsMargins(0, 0, 0, 0)
box = gui.vBox(self.mainArea, box=True)
sbox = gui.hBox(box)
gui.rubber(sbox)
gui.comboBox(sbox, self, "selected_quantity",
items=self.quantities, label="Show: ",
orientation=Qt.Horizontal, callback=self._update)
self.tablemodel = QStandardItemModel(self)
view = self.tableview = QTableView(
editTriggers=QTableView.NoEditTriggers)
view.setModel(self.tablemodel)
view.horizontalHeader().hide()
view.verticalHeader().hide()
view.horizontalHeader().setMinimumSectionSize(60)
view.selectionModel().selectionChanged.connect(self._invalidate)
view.setShowGrid(False)
view.setItemDelegate(BorderedItemDelegate(Qt.white))
view.setSizePolicy(QSizePolicy.MinimumExpanding,
QSizePolicy.MinimumExpanding)
view.clicked.connect(self.cell_clicked)
box.layout().addWidget(view)
selbox = gui.hBox(box)
gui.button(selbox, self, "Select Correct",
callback=self.select_correct, autoDefault=False)
gui.button(selbox, self, "Select Misclassified",
callback=self.select_wrong, autoDefault=False)
gui.button(selbox, self, "Clear Selection",
callback=self.select_none, autoDefault=False)
def sizeHint(self):
"""Initial size"""
return QSize(750, 340)
def _item(self, i, j):
return self.tablemodel.item(i, j) or QStandardItem()
def _set_item(self, i, j, item):
self.tablemodel.setItem(i, j, item)
def _init_table(self, nclasses):
item = self._item(0, 2)
item.setData("Predicted", Qt.DisplayRole)
item.setTextAlignment(Qt.AlignCenter)
item.setFlags(Qt.NoItemFlags)
self._set_item(0, 2, item)
item = self._item(2, 0)
item.setData("Actual", Qt.DisplayRole)
item.setTextAlignment(Qt.AlignHCenter | Qt.AlignBottom)
item.setFlags(Qt.NoItemFlags)
self.tableview.setItemDelegateForColumn(0, gui.VerticalItemDelegate())
self._set_item(2, 0, item)
self.tableview.setSpan(0, 2, 1, nclasses)
self.tableview.setSpan(2, 0, nclasses, 1)
font = self.tablemodel.invisibleRootItem().font()
bold_font = QFont(font)
bold_font.setBold(True)
for i in (0, 1):
for j in (0, 1):
item = self._item(i, j)
item.setFlags(Qt.NoItemFlags)
self._set_item(i, j, item)
for p, label in enumerate(self.headers):
for i, j in ((1, p + 2), (p + 2, 1)):
item = self._item(i, j)
item.setData(label, Qt.DisplayRole)
item.setFont(bold_font)
item.setTextAlignment(Qt.AlignRight | Qt.AlignVCenter)
item.setFlags(Qt.ItemIsEnabled)
if p < len(self.headers) - 1:
item.setData("br"[j == 1], BorderRole)
item.setData(QColor(192, 192, 192), BorderColorRole)
self._set_item(i, j, item)
hor_header = self.tableview.horizontalHeader()
if len(' '.join(self.headers)) < 120:
hor_header.setSectionResizeMode(QHeaderView.ResizeToContents)
else:
hor_header.setDefaultSectionSize(60)
self.tablemodel.setRowCount(nclasses + 3)
self.tablemodel.setColumnCount(nclasses + 3)
@Inputs.evaluation_results
def set_results(self, results):
"""Set the input results."""
prev_sel_learner = self.selected_learner.copy()
self.clear()
self.warning()
self.closeContext()
data = None
if results is not None and results.data is not None:
data = results.data[results.row_indices]
if data is not None and not data.domain.has_discrete_class:
self.Error.no_regression()
data = results = None
else:
self.Error.no_regression.clear()
nan_values = False
if results is not None:
assert isinstance(results, Orange.evaluation.Results)
if np.any(np.isnan(results.actual)) or \
np.any(np.isnan(results.predicted)):
# Error out here (could filter them out with a warning
# instead).
nan_values = True
results = data = None
if nan_values:
self.Error.invalid_values()
else:
self.Error.invalid_values.clear()
self.results = results
self.data = data
if data is not None:
class_values = data.domain.class_var.values
elif results is not None:
raise NotImplementedError
if results is None:
self.report_button.setDisabled(True)
else:
self.report_button.setDisabled(False)
nmodels = results.predicted.shape[0]
self.headers = class_values + \
[unicodedata.lookup("N-ARY SUMMATION")]
# NOTE: The 'learner_names' is set in 'Test Learners' widget.
if hasattr(results, "learner_names"):
self.learners = results.learner_names
else:
self.learners = ["Learner #{}".format(i + 1)
for i in range(nmodels)]
self._init_table(len(class_values))
self.openContext(data.domain.class_var)
if not prev_sel_learner or prev_sel_learner[0] >= len(self.learners):
if self.learners:
self.selected_learner[:] = [0]
else:
self.selected_learner[:] = prev_sel_learner
self._update()
self._set_selection()
self.unconditional_commit()
def clear(self):
"""Reset the widget, clear controls"""
self.results = None
self.data = None
self.tablemodel.clear()
self.headers = []
# Clear learners last. This action will invoke `_learner_changed`
self.learners = []
def select_correct(self):
"""Select the diagonal elements of the matrix"""
selection = QItemSelection()
n = self.tablemodel.rowCount()
for i in range(2, n):
index = self.tablemodel.index(i, i)
selection.select(index, index)
self.tableview.selectionModel().select(
selection, QItemSelectionModel.ClearAndSelect)
def select_wrong(self):
"""Select the off-diagonal elements of the matrix"""
selection = QItemSelection()
n = self.tablemodel.rowCount()
for i in range(2, n):
for j in range(i + 1, n):
index = self.tablemodel.index(i, j)
selection.select(index, index)
index = self.tablemodel.index(j, i)
selection.select(index, index)
self.tableview.selectionModel().select(
selection, QItemSelectionModel.ClearAndSelect)
def select_none(self):
"""Reset selection"""
self.tableview.selectionModel().clear()
def cell_clicked(self, model_index):
"""Handle cell click event"""
i, j = model_index.row(), model_index.column()
if not i or not j:
return
n = self.tablemodel.rowCount()
index = self.tablemodel.index
selection = None
if i == j == 1 or i == j == n - 1:
selection = QItemSelection(index(2, 2), index(n - 1, n - 1))
elif i in (1, n - 1):
selection = QItemSelection(index(2, j), index(n - 1, j))
elif j in (1, n - 1):
selection = QItemSelection(index(i, 2), index(i, n - 1))
if selection is not None:
self.tableview.selectionModel().select(
selection, QItemSelectionModel.ClearAndSelect)
def _prepare_data(self):
indices = self.tableview.selectedIndexes()
indices = {(ind.row() - 2, ind.column() - 2) for ind in indices}
actual = self.results.actual
learner_name = self.learners[self.selected_learner[0]]
predicted = self.results.predicted[self.selected_learner[0]]
selected = [i for i, t in enumerate(zip(actual, predicted))
if t in indices]
extra = []
class_var = self.data.domain.class_var
metas = self.data.domain.metas
if self.append_predictions:
extra.append(predicted.reshape(-1, 1))
var = Orange.data.DiscreteVariable(
"{}({})".format(class_var.name, learner_name),
class_var.values
)
metas = metas + (var,)
if self.append_probabilities and \
self.results.probabilities is not None:
probs = self.results.probabilities[self.selected_learner[0]]
extra.append(np.array(probs, dtype=object))
pvars = [Orange.data.ContinuousVariable("p({})".format(value))
for value in class_var.values]
metas = metas + tuple(pvars)
domain = Orange.data.Domain(self.data.domain.attributes,
self.data.domain.class_vars,
metas)
data = self.data.transform(domain)
if len(extra):
data.metas[:, len(self.data.domain.metas):] = \
np.hstack(tuple(extra))
data.name = learner_name
if selected:
annotated_data = create_annotated_table(data, selected)
data = data[selected]
else:
annotated_data = create_annotated_table(data, [])
data = None
return data, annotated_data
def commit(self):
"""Output data instances corresponding to selected cells"""
if self.results is not None and self.data is not None \
and self.selected_learner:
data, annotated_data = self._prepare_data()
else:
data = None
annotated_data = None
self.Outputs.selected_data.send(data)
self.Outputs.annotated_data.send(annotated_data)
def _invalidate(self):
indices = self.tableview.selectedIndexes()
self.selection = {(ind.row() - 2, ind.column() - 2) for ind in indices}
self.commit()
def _set_selection(self):
selection = QItemSelection()
index = self.tableview.model().index
for row, col in self.selection:
sel = index(row + 2, col + 2)
selection.select(sel, sel)
self.tableview.selectionModel().select(
selection, QItemSelectionModel.ClearAndSelect)
def _learner_changed(self):
self._update()
self._set_selection()
self.commit()
def _update(self):
def _isinvalid(x):
return isnan(x) or isinf(x)
# Update the displayed confusion matrix
if self.results is not None and self.selected_learner:
cmatrix = confusion_matrix(self.results, self.selected_learner[0])
colsum = cmatrix.sum(axis=0)
rowsum = cmatrix.sum(axis=1)
n = len(cmatrix)
diag = np.diag_indices(n)
colors = cmatrix.astype(np.double)
colors[diag] = 0
if self.selected_quantity == 0:
normalized = cmatrix.astype(np.int)
formatstr = "{}"
div = np.array([colors.max()])
else:
if self.selected_quantity == 1:
normalized = 100 * cmatrix / colsum
div = colors.max(axis=0)
else:
normalized = 100 * cmatrix / rowsum[:, np.newaxis]
div = colors.max(axis=1)[:, np.newaxis]
formatstr = "{:2.1f} %"
div[div == 0] = 1
colors /= div
colors[diag] = normalized[diag] / normalized[diag].max()
for i in range(n):
for j in range(n):
val = normalized[i, j]
col_val = colors[i, j]
item = self._item(i + 2, j + 2)
item.setData(
"NA" if _isinvalid(val) else formatstr.format(val),
Qt.DisplayRole)
bkcolor = QColor.fromHsl(
[0, 240][i == j], 160,
255 if _isinvalid(col_val) else int(255 - 30 * col_val))
item.setData(QBrush(bkcolor), Qt.BackgroundRole)
item.setData("trbl", BorderRole)
item.setToolTip("actual: {}\npredicted: {}".format(
self.headers[i], self.headers[j]))
item.setTextAlignment(Qt.AlignRight | Qt.AlignVCenter)
item.setFlags(Qt.ItemIsEnabled | Qt.ItemIsSelectable)
self._set_item(i + 2, j + 2, item)
bold_font = self.tablemodel.invisibleRootItem().font()
bold_font.setBold(True)
def _sum_item(value, border=""):
item = QStandardItem()
item.setData(value, Qt.DisplayRole)
item.setTextAlignment(Qt.AlignRight | Qt.AlignVCenter)
item.setFlags(Qt.ItemIsEnabled)
item.setFont(bold_font)
item.setData(border, BorderRole)
item.setData(QColor(192, 192, 192), BorderColorRole)
return item
for i in range(n):
self._set_item(n + 2, i + 2, _sum_item(int(colsum[i]), "t"))
self._set_item(i + 2, n + 2, _sum_item(int(rowsum[i]), "l"))
self._set_item(n + 2, n + 2, _sum_item(int(rowsum.sum())))
def send_report(self):
"""Send report"""
if self.results is not None and self.selected_learner:
self.report_table(
"Confusion matrix for {} (showing {})".
format(self.learners[self.selected_learner[0]],
self.quantities[self.selected_quantity].lower()),
self.tableview)
@classmethod
def migrate_settings(cls, settings, version):
if not version:
# For some period of time the 'selected_learner' property was
# changed from List[int] -> int
# (commit 4e49bb3fd0e11262f3ebf4b1116a91a4b49cc982) and then back
# again (commit 8a492d79a2e17154a0881e24a05843406c8892c0)
if "selected_learner" in settings and \
isinstance(settings["selected_learner"], int):
settings["selected_learner"] = [settings["selected_learner"]]
class OWKMeans(widget.OWWidget):
name = "k-Means"
description = "k-Means clustering algorithm with silhouette-based " \
"quality estimation."
icon = "icons/KMeans.svg"
priority = 2100
inputs = [("Data", Table, "set_data")]
outputs = [("Annotated Data", Table, widget.Default),
("Centroids", Table)]
INIT_KMEANS, INIT_RANDOM = range(2)
INIT_METHODS = "Initialize with KMeans++", "Random initialization"
SILHOUETTE, INTERCLUSTER, DISTANCES = range(3)
SCORING_METHODS = [("Silhouette", lambda km: km.silhouette, False),
("Inter-cluster distance",
lambda km: km.inter_cluster, True),
("Distance to centroids",
lambda km: km.inertia, True)]
OUTPUT_CLASS, OUTPUT_ATTRIBUTE, OUTPUT_META = range(3)
OUTPUT_METHODS = ("Class", "Feature", "Meta")
resizing_enabled = False
k = Setting(3)
k_from = Setting(2)
k_to = Setting(8)
optimize_k = Setting(False)
max_iterations = Setting(300)
n_init = Setting(10)
smart_init = Setting(INIT_KMEANS)
scoring = Setting(SILHOUETTE)
append_cluster_ids = Setting(True)
place_cluster_ids = Setting(OUTPUT_CLASS)
output_name = Setting("Cluster")
auto_run = Setting(True)
def __init__(self):
super().__init__()
self.data = None
self.km = None
self.optimization_runs = []
box = gui.vBox(self.controlArea, "Number of Clusters")
layout = QGridLayout()
self.n_clusters = bg = gui.radioButtonsInBox(
box, self, "optimize_k", [], orientation=layout,
callback=self.update)
layout.addWidget(
gui.appendRadioButton(bg, "Fixed:", addToLayout=False),
1, 1)
sb = gui.hBox(None, margin=0)
self.fixedSpinBox = gui.spin(
sb, self, "k", minv=2, maxv=30,
controlWidth=60, alignment=Qt.AlignRight, callback=self.update_k)
gui.rubber(sb)
layout.addWidget(sb, 1, 2)
layout.addWidget(
gui.appendRadioButton(bg, "Optimized from", addToLayout=False), 2, 1)
ftobox = gui.hBox(None)
ftobox.layout().setContentsMargins(0, 0, 0, 0)
layout.addWidget(ftobox)
gui.spin(
ftobox, self, "k_from", minv=2, maxv=29,
controlWidth=60, alignment=Qt.AlignRight,
callback=self.update_from)
gui.widgetLabel(ftobox, "to")
self.fixedSpinBox = gui.spin(
ftobox, self, "k_to", minv=3, maxv=30,
controlWidth=60, alignment=Qt.AlignRight,
callback=self.update_to)
gui.rubber(ftobox)
layout.addWidget(gui.widgetLabel(None, "Scoring: "),
5, 1, Qt.AlignRight)
layout.addWidget(
gui.comboBox(
None, self, "scoring", label="Scoring",
items=list(zip(*self.SCORING_METHODS))[0],
callback=self.update), 5, 2)
box = gui.vBox(self.controlArea, "Initialization")
gui.comboBox(
box, self, "smart_init", items=self.INIT_METHODS,
callback=self.update)
layout = QGridLayout()
box2 = gui.widgetBox(box, orientation=layout)
box2.setSizePolicy(QSizePolicy.Maximum, QSizePolicy.Maximum)
layout.addWidget(gui.widgetLabel(None, "Re-runs: "),
0, 0, Qt.AlignLeft)
sb = gui.hBox(None, margin=0)
layout.addWidget(sb, 0, 1)
gui.lineEdit(
sb, self, "n_init", controlWidth=60,
valueType=int, validator=QIntValidator(),
callback=self.update)
layout.addWidget(gui.widgetLabel(None, "Maximal iterations: "),
1, 0, Qt.AlignLeft)
sb = gui.hBox(None, margin=0)
layout.addWidget(sb, 1, 1)
gui.lineEdit(sb, self, "max_iterations",
controlWidth=60, valueType=int,
validator=QIntValidator(),
callback=self.update)
box = gui.vBox(self.controlArea, "Output")
gui.comboBox(box, self, "place_cluster_ids",
label="Append cluster ID as:", orientation=Qt.Horizontal,
callback=self.send_data, items=self.OUTPUT_METHODS)
gui.lineEdit(box, self, "output_name",
label="Name:", orientation=Qt.Horizontal,
callback=self.send_data)
gui.separator(self.buttonsArea, 30)
self.apply_button = gui.auto_commit(
self.buttonsArea, self, "auto_run", "Apply", box=None,
commit=self.commit
)
gui.rubber(self.controlArea)
self.table_model = QStandardItemModel(self)
self.table_model.setHorizontalHeaderLabels(["k", "Score"])
self.table_model.setColumnCount(2)
self.table_box = gui.vBox(
self.mainArea, "Optimization Report", addSpace=0)
table = self.table_view = QTableView(self.table_box)
table.setHorizontalScrollMode(QTableView.ScrollPerPixel)
table.setHorizontalScrollBarPolicy(Qt.ScrollBarAlwaysOff)
table.setSelectionMode(QTableView.SingleSelection)
table.setSelectionBehavior(QTableView.SelectRows)
table.verticalHeader().hide()
table.setItemDelegateForColumn(1, gui.TableBarItem(self))
table.setModel(self.table_model)
table.selectionModel().selectionChanged.connect(
self.table_item_selected)
table.setColumnWidth(0, 40)
table.setColumnWidth(1, 120)
table.horizontalHeader().setStretchLastSection(True)
self.setSizePolicy(QSizePolicy.Preferred, QSizePolicy.Preferred)
self.mainArea.setSizePolicy(QSizePolicy.Maximum,
QSizePolicy.Preferred)
self.table_box.setSizePolicy(QSizePolicy.Fixed,
QSizePolicy.MinimumExpanding)
self.table_view.setSizePolicy(QSizePolicy.Preferred,
QSizePolicy.MinimumExpanding)
self.table_box.layout().addWidget(self.table_view)
self.hide_show_opt_results()
def adjustSize(self):
self.ensurePolished()
s = self.sizeHint()
self.resize(s)
def hide_show_opt_results(self):
[self.mainArea.hide, self.mainArea.show][self.optimize_k]()
QTimer.singleShot(100, self.adjustSize)
def sizeHint(self):
s = self.controlArea.sizeHint()
if self.optimize_k and not self.mainArea.isHidden():
s.setWidth(s.width() + self.mainArea.sizeHint().width() +
4 * self.childrenRect().x())
return s
def update_k(self):
self.optimize_k = False
self.update()
def update_from(self):
self.k_to = max(self.k_from + 1, self.k_to)
self.optimize_k = True
self.update()
def update_to(self):
self.k_from = min(self.k_from, self.k_to - 1)
self.optimize_k = True
self.update()
def set_optimization(self):
self.updateOptimizationGui()
self.update()
def check_data_size(self, n, msg_group):
msg_group.add_message(
"not_enough_data",
"Too few ({}) unique data instances for {} clusters")
if n > len(self.data):
msg_group.not_enough_data(len(self.data), n)
return False
else:
msg_group.not_enough_data.clear()
return True
def run_optimization(self):
# Disabling is needed since this function is not reentrant
# Fast clicking on, say, "To: " causes multiple calls
try:
self.controlArea.setDisabled(True)
self.optimization_runs = []
if not self.check_data_size(self.k_from, self.Error):
return
self.check_data_size(self.k_to, self.Warning)
k_to = min(self.k_to, len(self.data))
kmeans = KMeans(
init=['random', 'k-means++'][self.smart_init],
n_init=self.n_init, max_iter=self.max_iterations,
compute_silhouette_score=self.scoring == self.SILHOUETTE)
with self.progressBar(k_to - self.k_from + 1) as progress:
for k in range(self.k_from, k_to + 1):
progress.advance()
kmeans.params["n_clusters"] = k
self.optimization_runs.append((k, kmeans(self.data)))
finally:
self.controlArea.setDisabled(False)
self.show_results()
self.send_data()
def cluster(self):
if not self.check_data_size(self.k, self.Error):
return
self.km = KMeans(
n_clusters=self.k,
init=['random', 'k-means++'][self.smart_init],
n_init=self.n_init,
max_iter=self.max_iterations)(self.data)
self.send_data()
def run(self):
self.clear_messages()
if not self.data:
return
if self.optimize_k:
self.run_optimization()
else:
self.cluster()
def commit(self):
self.run()
def show_results(self):
minimize = self.SCORING_METHODS[self.scoring][2]
k_scores = [(k, self.SCORING_METHODS[self.scoring][1](run)) for
k, run in self.optimization_runs]
scores = list(zip(*k_scores))[1]
if minimize:
best_score, worst_score = min(scores), max(scores)
else:
best_score, worst_score = max(scores), min(scores)
best_run = scores.index(best_score)
score_span = (best_score - worst_score) or 1
max_score = max(scores)
nplaces = min(5, np.floor(abs(math.log(max(max_score, 1e-10)))) + 2)
fmt = "{{:.{}f}}".format(int(nplaces))
model = self.table_model
model.setRowCount(len(k_scores))
for i, (k, score) in enumerate(k_scores):
item = model.item(i, 0)
if item is None:
item = QStandardItem()
item.setData(k, Qt.DisplayRole)
item.setTextAlignment(Qt.AlignCenter)
model.setItem(i, 0, item)
item = model.item(i, 1)
if item is None:
item = QStandardItem()
item.setData(fmt.format(score) if not np.isnan(score) else 'out-of-memory error',
Qt.DisplayRole)
bar_ratio = 0.95 * (score - worst_score) / score_span
item.setData(bar_ratio, gui.TableBarItem.BarRole)
model.setItem(i, 1, item)
self.table_view.resizeRowsToContents()
self.table_view.selectRow(best_run)
self.table_view.show()
if minimize:
self.table_box.setTitle("Scoring (smaller is better)")
else:
self.table_box.setTitle("Scoring (bigger is better)")
QTimer.singleShot(0, self.adjustSize)
def update(self):
self.hide_show_opt_results()
self.run()
def selected_row(self):
indices = self.table_view.selectedIndexes()
rows = {ind.row() for ind in indices}
if len(rows) == 1:
return rows.pop()
def table_item_selected(self):
row = self.selected_row()
if row is not None:
self.send_data(row)
def send_data(self, row=None):
if self.optimize_k:
if row is None:
row = self.selected_row()
km = self.optimization_runs[row][1]
else:
km = self.km
if not self.data or not km:
self.send("Annotated Data", None)
self.send("Centroids", None)
return
clust_var = DiscreteVariable(
self.output_name, values=["C%d" % (x + 1) for x in range(km.k)])
clust_ids = km(self.data)
domain = self.data.domain
attributes, classes = domain.attributes, domain.class_vars
meta_attrs = domain.metas
if self.place_cluster_ids == self.OUTPUT_CLASS:
if classes:
meta_attrs += classes
classes = [clust_var]
elif self.place_cluster_ids == self.OUTPUT_ATTRIBUTE:
attributes += (clust_var, )
else:
meta_attrs += (clust_var, )
domain = Domain(attributes, classes, meta_attrs)
new_table = Table.from_table(domain, self.data)
new_table.get_column_view(clust_var)[0][:] = clust_ids.X.ravel()
centroids = Table(Domain(km.pre_domain.attributes), km.centroids)
self.send("Annotated Data", new_table)
self.send("Centroids", centroids)
@check_sql_input
def set_data(self, data):
self.data = data
if data is None:
self.table_model.setRowCount(0)
self.send("Annotated Data", None)
self.send("Centroids", None)
else:
self.data = data
self.run()
def send_report(self):
k_clusters = self.k
if self.optimize_k and self.optimization_runs and self.selected_row() is not None:
k_clusters = self.optimization_runs[self.selected_row()][1].k
self.report_items((
("Number of clusters", k_clusters),
("Optimization",
self.optimize_k != 0 and
"{}, {} re-runs limited to {} steps".format(
self.INIT_METHODS[self.smart_init].lower(),
self.n_init, self.max_iterations)),
("Cluster ID in output",
self.append_cluster_ids and
"'{}' (as {})".format(
self.output_name,
self.OUTPUT_METHODS[self.place_cluster_ids].lower()))
))
if self.data:
self.report_data("Data", self.data)
if self.optimize_k:
self.report_table(
"Scoring by {}".format(self.SCORING_METHODS[self.scoring][0]
),
self.table_view)
