Box plot: Move 'order by importance' checkbox to groups

Author: janezd

Orange/widgets/visualize/owboxplot.py

@@ -203,7 +203,7 @@ def __init__(self):
         self.scene_min_x = self.scene_max_x = self.scene_width = 0
         self.label_width = 0
 
-        self.attrs = VariableListModel()
+        self.attrs = DomainModel(separators=False)
         view = gui.listView(
             self.controlArea, self, "attribute", box="Variable",
             model=self.attrs, callback=self.attr_changed)
@@ -212,18 +212,16 @@ def __init__(self):
         # set the minimal height (see the penultimate paragraph of
         # http://doc.qt.io/qt-4.8/qabstractscrollarea.html#addScrollBarWidget)
         view.setSizePolicy(QSizePolicy.Expanding, QSizePolicy.Ignored)
-        gui.separator(view.box, 6, 6)
+        self.group_vars = VariableListModel(placeholder="None")
+        self.group_view = view = gui.listView(
+            self.controlArea, self, "group_var", box="Subgroups",
+            model=self.group_vars, callback=self.grouping_changed)
+        gui.separator(self.group_view.box, 6, 6)
         self.cb_order = gui.checkBox(
             view.box, self, "order_by_importance",
             "Order by relevance",
             tooltip="Order by 𝜒² or ANOVA over the subgroups",
             callback=self.apply_sorting)
-        self.group_vars = DomainModel(
-            placeholder="None", separators=False,
-            valid_types=Orange.data.DiscreteVariable)
-        self.group_view = view = gui.listView(
-            self.controlArea, self, "group_var", box="Subgroups",
-            model=self.group_vars, callback=self.grouping_changed)
         view.setEnabled(False)
         view.setMinimumSize(QSize(30, 30))
         # See the comment above
@@ -289,11 +287,11 @@ def eventFilter(self, obj, event):
 
         return super().eventFilter(obj, event)
 
-    def reset_attrs(self, domain):
-        self.attrs[:] = [
+    def reset_groups(self, domain):
+        self.group_vars[:] = [
             var for var in chain(
                 domain.class_vars, domain.metas, domain.attributes)
-            if var.is_primitive()]
+            if var.is_discrete]
 
     # noinspection PyTypeChecker
     @Inputs.data
@@ -309,9 +307,9 @@ def set_data(self, dataset):
         self.attribute = None
         if dataset:
             domain = dataset.domain
-            self.group_vars.set_domain(domain)
+            self.attrs.set_domain(domain)
+            self.reset_groups(domain)
             self.group_view.setEnabled(len(self.group_vars) > 1)
-            self.reset_attrs(domain)
             self.select_default_variables(domain)
             self.openContext(self.dataset)
             self.grouping_changed()
@@ -332,37 +330,34 @@ def select_default_variables(self, domain):
             self.group_var = None  # Reset to trigger selection via callback
 
     def apply_sorting(self):
-        def compute_score(attr):
-            if attr is group_var:
+        def compute_anova(group):
+            group_col = data.get_column_view(group)[0].astype(float)
+            groups = (attr_col[group_col == i]
+                      for i in range(len(group.values)))
+            groups = (col[~np.isnan(col)] for col in groups)
+            groups = [group for group in groups if len(group)]
+            p = f_oneway(*groups)[1] if len(groups) > 1 else 2
+            return 2 if math.isnan(p) else p
+
+        def compute_chi(group):
+            if group is attr:
                 return 3
-            if attr.is_continuous:
-                # One-way ANOVA
-                col = data.get_column_view(attr)[0].astype(float)
-                groups = (col[group_col == i] for i in range(n_groups))
-                groups = (col[~np.isnan(col)] for col in groups)
-                groups = [group for group in groups if len(group)]
-                p = f_oneway(*groups)[1] if len(groups) > 1 else 2
-            else:
-                p = self._chi_square(group_var, attr)[1]
-            if math.isnan(p):
-                return 2
-            return p
+            p = self._chi_square(group, attr)[1]
+            return 2 if math.isnan(p) else p
 
         data = self.dataset
         if data is None:
             return
         domain = data.domain
-        attribute = self.attribute
-        group_var = self.group_var
-        if self.order_by_importance and group_var is not None:
-            n_groups = len(group_var.values)
-            group_col = data.get_column_view(group_var)[0] if \
-                domain.has_continuous_attributes(
-                    include_class=True, include_metas=True) else None
-            self.attrs.sort(key=compute_score)
+        attr = self.attribute
+        if self.order_by_importance:
+            if attr.is_continuous:
+                attr_col = data.get_column_view(attr)[0]
+                self.group_vars.sort(key=compute_anova)
+            else:
+                self.group_vars.sort(key=compute_chi)
         else:
-            self.reset_attrs(domain)
-        self.attribute = attribute
+            self.reset_groups(domain)
 
     def _chi_square(self, group_var, attr):
         # Chi-square with the given distribution into groups
@@ -379,15 +374,14 @@ def _chi_square(self, group_var, attr):
     def reset_all_data(self):
         self.clear_scene()
         self.stat_test = ""
-        self.attrs.clear()
-        self.group_vars.set_domain(None)
+        self.attrs.set_domain(None)
+        self.group_vars.clear()
         self.group_view.setEnabled(False)
         self.is_continuous = False
         self.update_display_box()
 
     def grouping_changed(self):
         self.cb_order.setEnabled(self.group_var is not None)
-        self.apply_sorting()
         self.attr_changed()
 
     def select_box_items(self):
@@ -399,6 +393,7 @@ def select_box_items(self):
 
     def attr_changed(self):
         self.compute_box_data()
+        self.apply_sorting()
         self.update_display_box()
         self.layout_changed()