class OWDiscretize(widget.OWWidget):
# pylint: disable=too-many-instance-attributes
name = "Discretize"
description = "Discretize the numeric data features."
icon = "icons/Discretize.svg"
keywords = ["bin", "categorical", "nominal", "ordinal"]
class Inputs:
data = Input("Data", Orange.data.Table, doc="Input data table")
class Outputs:
data = Output("Data", Orange.data.Table, doc="Table with discretized features")
settingsHandler = settings.DomainContextHandler()
settings_version = 2
saved_var_states = settings.ContextSetting({})
#: The default method name
default_method_name = settings.Setting(Methods.EqualFreq.name)
#: The k for Equal{Freq,Width}
default_k = settings.Setting(3)
#: The default cut points for custom entry
default_cutpoints: Tuple[float, ...] = settings.Setting(())
autosend = settings.Setting(True)
#: Discretization methods
Default, Leave, MDL, EqualFreq, EqualWidth, Remove, Custom = list(Methods)
want_main_area = False
resizing_enabled = False
def __init__(self):
super().__init__()
#: input data
self.data = None
self.class_var = None
#: Current variable discretization state
self.var_state = {}
#: Saved variable discretization settings (context setting)
self.saved_var_states = {}
self.method = Methods.Default
self.k = 5
self.cutpoints = ()
box = gui.vBox(self.controlArea, self.tr("Default Discretization"))
self._default_method_ = 0
self.default_bbox = rbox = gui.radioButtons(
box, self, "_default_method_", callback=self._default_disc_changed)
self.default_button_group = bg = rbox.findChild(QButtonGroup)
bg.buttonClicked[int].connect(self.set_default_method)
rb = gui.hBox(rbox)
self.left = gui.vBox(rb)
right = gui.vBox(rb)
rb.layout().setStretch(0, 1)
rb.layout().setStretch(1, 1)
self.options = [
(Methods.Default, self.tr("Default")),
(Methods.Leave, self.tr("Leave numeric")),
(Methods.MDL, self.tr("Entropy-MDL discretization")),
(Methods.EqualFreq, self.tr("Equal-frequency discretization")),
(Methods.EqualWidth, self.tr("Equal-width discretization")),
(Methods.Remove, self.tr("Remove numeric variables")),
(Methods.Custom, self.tr("Manual")),
]
for id_, opt in self.options[1:]:
t = gui.appendRadioButton(rbox, opt)
bg.setId(t, id_)
t.setChecked(id_ == self.default_method)
[right, self.left][opt.startswith("Equal")].layout().addWidget(t)
def _intbox(parent, attr, callback):
box = gui.indentedBox(parent)
s = gui.spin(
box, self, attr, minv=2, maxv=10, label="Num. of intervals:",
callback=callback)
s.setMaximumWidth(60)
s.setAlignment(Qt.AlignRight)
gui.rubber(s.box)
return box.box
self.k_general = _intbox(self.left, "default_k",
self._default_disc_changed)
self.k_general.layout().setContentsMargins(0, 0, 0, 0)
def manual_cut_editline(text="", enabled=True) -> QLineEdit:
edit = QLineEdit(
text=text,
placeholderText="e.g. 0.0, 0.5, 1.0",
toolTip="Enter fixed discretization cut points (a comma "
"separated list of strictly increasing numbers e.g. "
"0.0, 0.5, 1.0).",
enabled=enabled,
)
@edit.textChanged.connect
def update():
validator = edit.validator()
if validator is not None:
state, _, _ = validator.validate(edit.text(), 0)
else:
state = QValidator.Acceptable
palette = edit.palette()
colors = {
QValidator.Intermediate: (Qt.yellow, Qt.black),
QValidator.Invalid: (Qt.red, Qt.black),
}.get(state, None)
if colors is None:
palette = QPalette()
else:
palette.setColor(QPalette.Base, colors[0])
palette.setColor(QPalette.Text, colors[1])
cr = edit.cursorRect()
p = edit.mapToGlobal(cr.bottomRight())
edit.setPalette(palette)
if state != QValidator.Acceptable and edit.isVisible():
show_tip(edit, p, edit.toolTip(), textFormat=Qt.RichText)
else:
show_tip(edit, p, "")
return edit
self.manual_cuts_edit = manual_cut_editline(
text=", ".join(map(str, self.default_cutpoints)),
enabled=self.default_method == Methods.Custom,
)
def set_manual_default_cuts():
text = self.manual_cuts_edit.text()
self.default_cutpoints = tuple(
float(s.strip()) for s in text.split(",") if s.strip())
self._default_disc_changed()
self.manual_cuts_edit.editingFinished.connect(set_manual_default_cuts)
validator = IncreasingNumbersListValidator()
self.manual_cuts_edit.setValidator(validator)
ibox = gui.indentedBox(right, orientation=Qt.Horizontal)
ibox.layout().addWidget(self.manual_cuts_edit)
right.layout().addStretch(10)
self.left.layout().addStretch(10)
self.connect_control(
"default_cutpoints",
lambda values: self.manual_cuts_edit.setText(", ".join(map(str, values)))
)
vlayout = QHBoxLayout()
box = gui.widgetBox(
self.controlArea, "Individual Attribute Settings",
orientation=vlayout, spacing=8
)
# List view with all attributes
self.varview = ListViewSearch(
selectionMode=QListView.ExtendedSelection,
uniformItemSizes=True,
)
self.varview.setItemDelegate(DiscDelegate())
self.varmodel = itemmodels.VariableListModel()
self.varview.setModel(self.varmodel)
self.varview.selectionModel().selectionChanged.connect(
self._var_selection_changed
)
vlayout.addWidget(self.varview)
# Controls for individual attr settings
self.bbox = controlbox = gui.radioButtons(
box, self, "method", callback=self._disc_method_changed
)
vlayout.addWidget(controlbox)
self.variable_button_group = bg = controlbox.findChild(QButtonGroup)
for id_, opt in self.options[:5]:
b = gui.appendRadioButton(controlbox, opt)
bg.setId(b, id_)
self.k_specific = _intbox(controlbox, "k", self._disc_method_changed)
gui.appendRadioButton(controlbox, "Remove attribute", id=Methods.Remove)
b = gui.appendRadioButton(controlbox, "Manual", id=Methods.Custom)
self.manual_cuts_specific = manual_cut_editline(
text=", ".join(map(str, self.cutpoints)),
enabled=self.method == Methods.Custom
)
self.manual_cuts_specific.setValidator(validator)
b.toggled[bool].connect(self.manual_cuts_specific.setEnabled)
def set_manual_cuts():
text = self.manual_cuts_specific.text()
points = [t for t in text.split(",") if t.split()]
self.cutpoints = tuple(float(t) for t in points)
self._disc_method_changed()
self.manual_cuts_specific.editingFinished.connect(set_manual_cuts)
self.connect_control(
"cutpoints",
lambda values: self.manual_cuts_specific.setText(", ".join(map(str, values)))
)
ibox = gui.indentedBox(controlbox, orientation=Qt.Horizontal)
self.copy_current_to_manual_button = b = FixedSizeButton(
text="CC", toolTip="Copy the current cut points to manual mode",
enabled=False
)
b.clicked.connect(self._copy_to_manual)
ibox.layout().addWidget(self.manual_cuts_specific)
ibox.layout().addWidget(b)
gui.rubber(controlbox)
controlbox.setEnabled(False)
bg.button(self.method)
self.controlbox = controlbox
gui.auto_apply(self.buttonsArea, self, "autosend")
self._update_spin_positions()
@property
def default_method(self) -> Methods:
return Methods[self.default_method_name]
@default_method.setter
def default_method(self, method):
self.set_default_method(method)
def set_default_method(self, method: Methods):
if isinstance(method, int):
method = Methods(method)
else:
method = Methods.from_method(method)
if method != self.default_method:
self.default_method_name = method.name
self.default_button_group.button(method).setChecked(True)
self._default_disc_changed()
self.manual_cuts_edit.setEnabled(method == Methods.Custom)
@Inputs.data
def set_data(self, data):
self.closeContext()
self.data = data
if self.data is not None:
self._initialize(data)
self.openContext(data)
# Restore the per variable discretization settings
self._restore(self.saved_var_states)
# Complete the induction of cut points
self._update_points()
else:
self._clear()
self.commit.now()
def _initialize(self, data):
# Initialize the default variable states for new data.
self.class_var = data.domain.class_var
cvars = [var for var in data.domain.variables
if var.is_continuous]
self.varmodel[:] = cvars
has_disc_class = data.domain.has_discrete_class
def set_enabled(box: QWidget, id_: Methods, state: bool):
bg = box.findChild(QButtonGroup)
b = bg.button(id_)
b.setEnabled(state)
set_enabled(self.default_bbox, self.MDL, has_disc_class)
bg = self.bbox.findChild(QButtonGroup)
b = bg.button(Methods.MDL)
b.setEnabled(has_disc_class)
set_enabled(self.bbox, self.MDL, has_disc_class)
# If the newly disabled MDL button is checked then change it
if not has_disc_class and self.default_method == self.MDL:
self.default_method = Methods.Leave
if not has_disc_class and self.method == self.MDL:
self.method = Methods.Default
# Reset (initialize) the variable discretization states.
self._reset()
def _restore(self, saved_state):
# Restore variable states from a saved_state dictionary.
def_method = self._current_default_method()
for i, var in enumerate(self.varmodel):
key = variable_key(var)
if key in saved_state:
state = saved_state[key]
if isinstance(state.method, Default):
state = DState(Default(def_method), None, None)
self._set_var_state(i, state)
def _reset(self):
# restore the individual variable settings back to defaults.
def_method = self._current_default_method()
self.var_state = {}
for i in range(len(self.varmodel)):
state = DState(Default(def_method), None, None)
self._set_var_state(i, state)
def _set_var_state(self, index, state):
# set the state of variable at `index` to `state`.
self.var_state[index] = state
self.varmodel.setData(self.varmodel.index(index), state, Qt.UserRole)
def _clear(self):
self.data = None
self.varmodel[:] = []
self.var_state = {}
self.saved_var_states = {}
self.default_button_group.button(self.MDL).setEnabled(True)
self.variable_button_group.button(self.MDL).setEnabled(True)
def _update_points(self):
"""
Update the induced cut points.
"""
if self.data is None:
return
def induce_cuts(method, data, var):
dvar = _dispatch[type(method)](method, data, var)
if dvar is None:
# removed
return [], None
elif dvar is var:
# no transformation took place
return None, var
elif is_discretized(dvar):
return dvar.compute_value.points, dvar
raise ValueError
for i, var in enumerate(self.varmodel):
state = self.var_state[i]
if state.points is None and state.disc_var is None:
points, dvar = induce_cuts(state.method, self.data, var)
new_state = state._replace(points=points, disc_var=dvar)
self._set_var_state(i, new_state)
def _current_default_method(self):
method = self.default_method
k = self.default_k
if method == Methods.Leave:
def_method = Leave()
elif method == Methods.MDL:
def_method = MDL()
elif method == Methods.EqualFreq:
def_method = EqualFreq(k)
elif method == Methods.EqualWidth:
def_method = EqualWidth(k)
elif method == Methods.Remove:
def_method = Remove()
elif method == Methods.Custom:
def_method = Custom(self.default_cutpoints)
else:
assert False
return def_method
def _current_method(self):
if self.method == Methods.Default:
method = Default(self._current_default_method())
elif self.method == Methods.Leave:
method = Leave()
elif self.method == Methods.MDL:
method = MDL()
elif self.method == Methods.EqualFreq:
method = EqualFreq(self.k)
elif self.method == Methods.EqualWidth:
method = EqualWidth(self.k)
elif self.method == Methods.Remove:
method = Remove()
elif self.method == Methods.Custom:
method = Custom(self.cutpoints)
else:
assert False
return method
def _update_spin_positions(self):
kmethods = [Methods.EqualFreq, Methods.EqualWidth]
self.k_general.setDisabled(self.default_method not in kmethods)
if self.default_method == Methods.EqualFreq:
self.left.layout().insertWidget(1, self.k_general)
elif self.default_method == Methods.EqualWidth:
self.left.layout().insertWidget(2, self.k_general)
self.k_specific.setDisabled(self.method not in kmethods)
if self.method == Methods.EqualFreq:
self.bbox.layout().insertWidget(4, self.k_specific)
elif self.method == Methods.EqualWidth:
self.bbox.layout().insertWidget(5, self.k_specific)
def _default_disc_changed(self):
self._update_spin_positions()
method = self._current_default_method()
state = DState(Default(method), None, None)
for i, _ in enumerate(self.varmodel):
if isinstance(self.var_state[i].method, Default):
self._set_var_state(i, state)
self._update_points()
self.commit.deferred()
def _disc_method_changed(self):
self._update_spin_positions()
indices = self.selected_indices()
method = self._current_method()
state = DState(method, None, None)
for idx in indices:
self._set_var_state(idx, state)
self._update_points()
self._copy_to_manual_update_enabled()
self.commit.deferred()
def _copy_to_manual(self):
indices = self.selected_indices()
# set of all methods for the current selection
if len(indices) != 1:
return
index = indices[0]
state = self.var_state[index]
var = self.varmodel[index]
fmt = var.repr_val
points = state.points
if points is None:
points = ()
else:
points = tuple(state.points)
state = state._replace(method=Custom(points), points=None, disc_var=None)
self._set_var_state(index, state)
self.method = Methods.Custom
self.cutpoints = points
self.manual_cuts_specific.setText(", ".join(map(fmt, points)))
self._update_points()
self.commit.deferred()
def _copy_to_manual_update_enabled(self):
indices = self.selected_indices()
methods = [self.var_state[i].method for i in indices]
self.copy_current_to_manual_button.setEnabled(
len(indices) == 1 and not isinstance(methods[0], Custom))
def _var_selection_changed(self, *_):
self._copy_to_manual_update_enabled()
indices = self.selected_indices()
# set of all methods for the current selection
methods = [self.var_state[i].method for i in indices]
def key(method):
if isinstance(method, Default):
return Default, (None, )
return type(method), tuple(method)
mset = list(unique_everseen(methods, key=key))
self.controlbox.setEnabled(len(mset) > 0)
if len(mset) == 1:
method = mset.pop()
self.method = Methods.from_method(method)
if isinstance(method, (EqualFreq, EqualWidth)):
self.k = method.k
elif isinstance(method, Custom):
self.cutpoints = method.points
else:
# deselect the current button
self.method = -1
bg = self.controlbox.group
button_group_reset(bg)
self._update_spin_positions()
def selected_indices(self):
rows = self.varview.selectionModel().selectedRows()
return [index.row() for index in rows]
def method_for_index(self, index):
state = self.var_state[index]
return state.method
def discretized_var(self, index):
# type: (int) -> Optional[Orange.data.DiscreteVariable]
state = self.var_state[index]
if state.disc_var is not None and state.points == []:
# Removed by MDL Entropy
return None
else:
return state.disc_var
def discretized_domain(self):
"""
Return the current effective discretized domain.
"""
if self.data is None:
return None
# a mapping of all applied changes for variables in `varmodel`
mapping = {var: self.discretized_var(i)
for i, var in enumerate(self.varmodel)}
def disc_var(source):
return mapping.get(source, source)
# map the full input domain to the new variables (where applicable)
attributes = [disc_var(v) for v in self.data.domain.attributes]
attributes = [v for v in attributes if v is not None]
class_vars = [disc_var(v) for v in self.data.domain.class_vars]
class_vars = [v for v in class_vars if v is not None]
domain = Orange.data.Domain(
attributes, class_vars, metas=self.data.domain.metas
)
return domain
@gui.deferred
def commit(self):
output = None
if self.data is not None:
domain = self.discretized_domain()
output = self.data.transform(domain)
self.Outputs.data.send(output)
def storeSpecificSettings(self):
super().storeSpecificSettings()
self.saved_var_states = {
variable_key(var):
self.var_state[i]._replace(points=None, disc_var=None)
for i, var in enumerate(self.varmodel)
}
def send_report(self):
self.report_items((
("Default method", self.options[self.default_method][1]),))
if self.varmodel:
self.report_items("Thresholds", [
(var.name,
DiscDelegate.cutsText(self.var_state[i], var.repr_val) or "leave numeric")
for i, var in enumerate(self.varmodel)])
@classmethod
def migrate_settings(cls, settings, version): # pylint: disable=redefined-outer-name
if version is None or version < 2:
# was stored as int indexing Methods (but offset by 1)
default = settings.pop("default_method", 0)
default = Methods(default + 1)
settings["default_method_name"] = default.name
class OWImpute(OWWidget):
name = "Impute"
description = "Impute missing values in the data table."
icon = "icons/Impute.svg"
priority = 2130
keywords = ["substitute", "missing"]
class Inputs:
data = Input("Data", Orange.data.Table)
learner = Input("Learner", Learner)
class Outputs:
data = Output("Data", Orange.data.Table)
class Error(OWWidget.Error):
imputation_failed = Msg("Imputation failed for '{}'")
model_based_imputer_sparse = \
Msg("Model based imputer does not work for sparse data")
class Warning(OWWidget.Warning):
cant_handle_var = Msg("Default method can not handle '{}'")
settingsHandler = settings.DomainContextHandler()
_default_method_index = settings.Setting(int(Method.Leave)) # type: int
# Per-variable imputation state (synced in storeSpecificSettings)
_variable_imputation_state = settings.ContextSetting(
{}) # type: VariableState
autocommit = settings.Setting(True)
want_main_area = False
resizing_enabled = False
def __init__(self):
super().__init__()
self.data = None # type: Optional[Orange.data.Table]
self.learner = None # type: Optional[Learner]
self.default_learner = SimpleTreeLearner(min_instances=10,
max_depth=10)
self.modified = False
self.executor = qconcurrent.ThreadExecutor(self)
self.__task = None
main_layout = QVBoxLayout()
main_layout.setContentsMargins(10, 10, 10, 10)
self.controlArea.layout().addLayout(main_layout)
box = QGroupBox(title=self.tr("Default Method"), flat=False)
box_layout = QGridLayout(box)
box_layout.setContentsMargins(5, 0, 0, 0)
main_layout.addWidget(box)
button_group = QButtonGroup()
button_group.buttonClicked[int].connect(self.set_default_method)
for i, (method, _) in enumerate(list(METHODS.items())[1:-1]):
imputer = self.create_imputer(method)
button = QRadioButton(imputer.name)
button.setChecked(method == self.default_method_index)
button_group.addButton(button, method)
box_layout.addWidget(button, i % 3, i // 3)
self.default_button_group = button_group
box = QGroupBox(title=self.tr("Individual Attribute Settings"),
flat=False)
main_layout.addWidget(box)
horizontal_layout = QHBoxLayout(box)
main_layout.addWidget(box)
self.varview = ListViewSearch(
selectionMode=QListView.ExtendedSelection, uniformItemSizes=True)
self.varview.setItemDelegate(DisplayFormatDelegate())
self.varmodel = itemmodels.VariableListModel()
self.varview.setModel(self.varmodel)
self.varview.selectionModel().selectionChanged.connect(
self._on_var_selection_changed)
self.selection = self.varview.selectionModel()
horizontal_layout.addWidget(self.varview)
method_layout = QVBoxLayout()
horizontal_layout.addLayout(method_layout)
button_group = QButtonGroup()
for method in Method:
imputer = self.create_imputer(method)
button = QRadioButton(text=imputer.name)
button_group.addButton(button, method)
method_layout.addWidget(button)
self.value_combo = QComboBox(
minimumContentsLength=8,
sizeAdjustPolicy=QComboBox.AdjustToMinimumContentsLength,
activated=self._on_value_selected)
self.value_double = QDoubleSpinBox(
editingFinished=self._on_value_selected,
minimum=-1000.,
maximum=1000.,
singleStep=.1,
decimals=3,
)
self.value_stack = value_stack = QStackedWidget()
value_stack.addWidget(self.value_combo)
value_stack.addWidget(self.value_double)
method_layout.addWidget(value_stack)
button_group.buttonClicked[int].connect(
self.set_method_for_current_selection)
method_layout.addStretch(2)
reset_button = QPushButton("Restore All to Default",
checked=False,
checkable=False,
clicked=self.reset_variable_state,
default=False,
autoDefault=False)
method_layout.addWidget(reset_button)
self.variable_button_group = button_group
box = gui.auto_apply(self.controlArea, self, "autocommit")
box.button.setFixedWidth(180)
box.layout().insertStretch(0)
self.info.set_input_summary(self.info.NoInput)
self.info.set_output_summary(self.info.NoOutput)
def create_imputer(self, method, *args):
# type: (Method, ...) -> impute.BaseImputeMethod
if method == Method.Model:
if self.learner is not None:
return impute.Model(self.learner)
else:
return impute.Model(self.default_learner)
elif method == Method.AsAboveSoBelow:
assert self.default_method_index != Method.AsAboveSoBelow
default = self.create_imputer(Method(self.default_method_index))
m = AsDefault()
m.method = default
return m
else:
return METHODS[method](*args)
@property
def default_method_index(self):
return self._default_method_index
@default_method_index.setter
def default_method_index(self, index):
if self._default_method_index != index:
assert index != Method.AsAboveSoBelow
self._default_method_index = index
self.default_button_group.button(index).setChecked(True)
# update variable view
self.update_varview()
self._invalidate()
def set_default_method(self, index):
"""Set the current selected default imputation method.
"""
self.default_method_index = index
@Inputs.data
@check_sql_input
def set_data(self, data):
self.cancel()
self.closeContext()
self.varmodel[:] = []
self._variable_imputation_state = {} # type: VariableState
self.modified = False
self.data = data
if data is not None:
self.varmodel[:] = data.domain.variables
self.openContext(data.domain)
# restore per variable imputation state
self._restore_state(self._variable_imputation_state)
summary = len(data) if data else self.info.NoInput
details = format_summary_details(data) if data else ""
self.info.set_input_summary(summary, details)
self.update_varview()
self.unconditional_commit()
@Inputs.learner
def set_learner(self, learner):
self.cancel()
self.learner = learner or self.default_learner
imputer = self.create_imputer(Method.Model)
button = self.default_button_group.button(Method.Model)
button.setText(imputer.name)
variable_button = self.variable_button_group.button(Method.Model)
variable_button.setText(imputer.name)
if learner is not None:
self.default_method_index = Method.Model
self.update_varview()
self.commit()
def get_method_for_column(self, column_index):
# type: (int) -> impute.BaseImputeMethod
"""
Return the imputation method for column by its index.
"""
assert 0 <= column_index < len(self.varmodel)
idx = self.varmodel.index(column_index, 0)
state = idx.data(StateRole)
if state is None:
state = (Method.AsAboveSoBelow, ())
return self.create_imputer(state[0], *state[1])
def _invalidate(self):
self.modified = True
if self.__task is not None:
self.cancel()
self.commit()
def commit(self):
self.cancel()
self.warning()
self.Error.imputation_failed.clear()
self.Error.model_based_imputer_sparse.clear()
summary = len(self.data) if self.data else self.info.NoOutput
detail = format_summary_details(self.data) if self.data else ""
self.info.set_output_summary(summary, detail)
if not self.data or not self.varmodel.rowCount():
self.Outputs.data.send(self.data)
self.modified = False
return
data = self.data
impute_state = [(i, var, self.get_method_for_column(i))
for i, var in enumerate(self.varmodel)]
# normalize to the effective method bypasing AsDefault
impute_state = [(i, var, m.method if isinstance(m, AsDefault) else m)
for i, var, m in impute_state]
def impute_one(method, var, data):
# Readability counts, pylint: disable=no-else-raise
# type: (impute.BaseImputeMethod, Variable, Table) -> Any
if isinstance(method, impute.Model) and data.is_sparse():
raise SparseNotSupported()
elif isinstance(method, impute.DropInstances):
return RowMask(method(data, var))
elif not method.supports_variable(var):
raise VariableNotSupported(var)
else:
return method(data, var)
futures = []
for _, var, method in impute_state:
f = self.executor.submit(impute_one, copy.deepcopy(method), var,
data)
futures.append(f)
w = qconcurrent.FutureSetWatcher(futures)
w.doneAll.connect(self.__commit_finish)
w.progressChanged.connect(self.__progress_changed)
self.__task = Task(futures, w)
self.progressBarInit()
self.setInvalidated(True)
@Slot()
def __commit_finish(self):
assert QThread.currentThread() is self.thread()
assert self.__task is not None
futures = self.__task.futures
assert len(futures) == len(self.varmodel)
assert self.data is not None
def get_variable(variable, future, drop_mask) \
-> Optional[List[Orange.data.Variable]]:
# Returns a (potentially empty) list of variables,
# or None on failure that should interrupt the imputation
assert future.done()
try:
res = future.result()
except SparseNotSupported:
self.Error.model_based_imputer_sparse()
return [] # None?
except VariableNotSupported:
self.Warning.cant_handle_var(variable.name)
return []
except Exception: # pylint: disable=broad-except
log = logging.getLogger(__name__)
log.info("Error for %s", variable.name, exc_info=True)
self.Error.imputation_failed(variable.name)
return None
if isinstance(res, RowMask):
drop_mask |= res.mask
newvar = variable
else:
newvar = res
if isinstance(newvar, Orange.data.Variable):
newvar = [newvar]
return newvar
def create_data(attributes, class_vars):
domain = Orange.data.Domain(attributes, class_vars,
self.data.domain.metas)
try:
return self.data.from_table(domain, self.data[~drop_mask])
except Exception: # pylint: disable=broad-except
log = logging.getLogger(__name__)
log.info("Error", exc_info=True)
self.Error.imputation_failed("Unknown")
return None
self.__task = None
self.setInvalidated(False)
self.progressBarFinished()
attributes = []
class_vars = []
drop_mask = np.zeros(len(self.data), bool)
for i, (var, fut) in enumerate(zip(self.varmodel, futures)):
newvar = get_variable(var, fut, drop_mask)
if newvar is None:
data = None
break
if i < len(self.data.domain.attributes):
attributes.extend(newvar)
else:
class_vars.extend(newvar)
else:
data = create_data(attributes, class_vars)
self.Outputs.data.send(data)
self.modified = False
summary = len(data) if data else self.info.NoOutput
details = format_summary_details(data) if data else ""
self.info.set_output_summary(summary, details)
@Slot(int, int)
def __progress_changed(self, n, d):
assert QThread.currentThread() is self.thread()
assert self.__task is not None
self.progressBarSet(100. * n / d)
def cancel(self):
self.__cancel(wait=False)
def __cancel(self, wait=False):
if self.__task is not None:
task, self.__task = self.__task, None
task.cancel()
task.watcher.doneAll.disconnect(self.__commit_finish)
task.watcher.progressChanged.disconnect(self.__progress_changed)
if wait:
concurrent.futures.wait(task.futures)
task.watcher.flush()
self.progressBarFinished()
self.setInvalidated(False)
def onDeleteWidget(self):
self.__cancel(wait=True)
super().onDeleteWidget()
def send_report(self):
specific = []
for i, var in enumerate(self.varmodel):
method = self.get_method_for_column(i)
if not isinstance(method, AsDefault):
specific.append("{} ({})".format(var.name, str(method)))
default = self.create_imputer(Method.AsAboveSoBelow)
if specific:
self.report_items((("Default method", default.name),
("Specific imputers", ", ".join(specific))))
else:
self.report_items((("Method", default.name), ))
def _on_var_selection_changed(self):
# Method is well documented, splitting it is not needed for readability,
# thus pylint: disable=too-many-branches
indexes = self.selection.selectedIndexes()
defmethod = (Method.AsAboveSoBelow, ())
methods = [index.data(StateRole) for index in indexes]
methods = [m if m is not None else defmethod for m in methods]
methods = set(methods)
selected_vars = [self.varmodel[index.row()] for index in indexes]
has_discrete = any(var.is_discrete for var in selected_vars)
fixed_value = None
value_stack_enabled = False
current_value_widget = None
if len(methods) == 1:
method_type, parameters = methods.pop()
for m in Method:
if method_type == m:
self.variable_button_group.button(m).setChecked(True)
if method_type == Method.Default:
(fixed_value, ) = parameters
elif self.variable_button_group.checkedButton() is not None:
# Uncheck the current button
self.variable_button_group.setExclusive(False)
self.variable_button_group.checkedButton().setChecked(False)
self.variable_button_group.setExclusive(True)
assert self.variable_button_group.checkedButton() is None
# Update variable methods GUI enabled state based on selection.
for method in Method:
# use a default constructed imputer to query support
imputer = self.create_imputer(method)
enabled = all(
imputer.supports_variable(var) for var in selected_vars)
button = self.variable_button_group.button(method)
button.setEnabled(enabled)
# Update the "Value" edit GUI.
if not has_discrete:
# no discrete variables -> allow mass edit for all (continuous vars)
value_stack_enabled = True
current_value_widget = self.value_double
elif len(selected_vars) == 1:
# single discrete var -> enable and fill the values combo
value_stack_enabled = True
current_value_widget = self.value_combo
self.value_combo.clear()
self.value_combo.addItems(selected_vars[0].values)
else:
# mixed type selection -> disable
value_stack_enabled = False
current_value_widget = None
self.variable_button_group.button(Method.Default).setEnabled(False)
self.value_stack.setEnabled(value_stack_enabled)
if current_value_widget is not None:
self.value_stack.setCurrentWidget(current_value_widget)
if fixed_value is not None:
# set current value
if current_value_widget is self.value_combo:
self.value_combo.setCurrentIndex(fixed_value)
elif current_value_widget is self.value_double:
self.value_double.setValue(fixed_value)
else:
assert False
def set_method_for_current_selection(self, method_index):
# type: (Method) -> None
indexes = self.selection.selectedIndexes()
self.set_method_for_indexes(indexes, method_index)
def set_method_for_indexes(self, indexes, method_index):
# type: (List[QModelIndex], Method) -> None
if method_index == Method.AsAboveSoBelow:
for index in indexes:
self.varmodel.setData(index, None, StateRole)
elif method_index == Method.Default:
current = self.value_stack.currentWidget()
if current is self.value_combo:
value = self.value_combo.currentIndex()
else:
value = self.value_double.value()
for index in indexes:
state = (int(Method.Default), (value, ))
self.varmodel.setData(index, state, StateRole)
else:
state = (int(method_index), ())
for index in indexes:
self.varmodel.setData(index, state, StateRole)
self.update_varview(indexes)
self._invalidate()
def update_varview(self, indexes=None):
if indexes is None:
indexes = map(self.varmodel.index, range(len(self.varmodel)))
for index in indexes:
self.varmodel.setData(index,
self.get_method_for_column(index.row()),
DisplayMethodRole)
def _on_value_selected(self):
# The fixed 'Value' in the widget has been changed by the user.
self.variable_button_group.button(Method.Default).setChecked(True)
self.set_method_for_current_selection(Method.Default)
def reset_variable_state(self):
indexes = list(map(self.varmodel.index, range(len(self.varmodel))))
self.set_method_for_indexes(indexes, Method.AsAboveSoBelow)
self.variable_button_group.button(
Method.AsAboveSoBelow).setChecked(True)
def _store_state(self):
# type: () -> VariableState
"""
Save the current variable imputation state
"""
state = {} # type: VariableState
for i, var in enumerate(self.varmodel):
index = self.varmodel.index(i)
m = index.data(StateRole)
if m is not None:
state[var_key(var)] = m
return state
def _restore_state(self, state):
# type: (VariableState) -> None
"""
Restore the variable imputation state from the saved state
"""
def check(state):
# check if state is a proper State
if isinstance(state, tuple) and len(state) == 2:
m, p = state
if isinstance(m, int) and isinstance(p, tuple) and \
0 <= m < len(Method):
return True
return False
for i, var in enumerate(self.varmodel):
m = state.get(var_key(var), None)
if check(m):
self.varmodel.setData(self.varmodel.index(i), m, StateRole)
def storeSpecificSettings(self):
self._variable_imputation_state = self._store_state()
super().storeSpecificSettings()