Add Cochran rule indicator to Sieve + tests

Orange/widgets/visualize/owsieve.py

@@ -8,7 +8,7 @@
 from AnyQt.QtGui import QColor, QPen, QBrush
 from AnyQt.QtWidgets import QGraphicsScene, QGraphicsLineItem, QSizePolicy
 
-from Orange.data import Table, filter, Variable
+from Orange.data import Table, filter as data_filter, Variable
 from Orange.data.sql.table import SqlTable, LARGE_TABLE, DEFAULT_SAMPLE_TIME
 from Orange.preprocess import Discretize
 from Orange.preprocess.discretize import EqualFreq
@@ -24,7 +24,8 @@
     VizRankMixin
 from Orange.widgets.visualize.utils import (
     CanvasText, CanvasRectangle, ViewWithPress)
-from Orange.widgets.widget import OWWidget, AttributeList, Input, Output
+from Orange.widgets.widget import OWWidget, AttributeList, Input, Output, \
+    Msg
 
 
 class ChiSqStats:
@@ -54,7 +55,6 @@ def __init__(self, data, attr1, attr2):
         self.p = chi2.sf(
             self.chisq, (len(self.probs_x) - 1) * (len(self.probs_y) - 1))
 
-
 class SieveRank(VizRankDialogAttrPair):
     sort_names_in_row = True
 
@@ -84,6 +84,12 @@ class Outputs:
 
     graph_name = "canvas"  # QGraphicsScene
 
+    class Information(OWWidget.Information):
+        cochran = Msg("Meets Cochran's rule")
+
+    class Warning(OWWidget.Warning):
+        cochran = Msg("Does not meet Cochran's rule")
+
     want_control_area = False
 
     settings_version = 1
@@ -105,10 +111,10 @@ def __init__(self):
         self.mainArea.layout().setSpacing(0)
         self.attr_box = gui.hBox(self.mainArea, margin=0)
         self.domain_model = DomainModel(valid_types=DomainModel.PRIMITIVE)
-        combo_args = dict(
-            widget=self.attr_box, master=self, contentsLength=12,
-            searchable=True, sendSelectedValue=True,
-            callback=self.attr_changed, model=self.domain_model)
+        combo_args = {
+            "widget":self.attr_box, "master":self, "contentsLength":12,
+            "searchable":True, "sendSelectedValue":True,
+            "callback":self.attr_changed, "model":self.domain_model}
         fixed_size = (QSizePolicy.Fixed, QSizePolicy.Fixed)
         gui.comboBox(value="attr_x", **combo_args)
         gui.widgetLabel(self.attr_box, "\u2717", sizePolicy=fixed_size)
@@ -270,7 +276,7 @@ def resolve_shown_attributes(self):
                 "Features from the input signal are not present in the data")
             return
         old_attrs = self.attr_x, self.attr_y
-        self.attr_x, self.attr_y = [f for f in (features * 2)[:2]]
+        self.attr_x, self.attr_y = (features * 2)[:2]
         self.attr_box.setEnabled(False)
         if (self.attr_x, self.attr_y) != old_attrs:
             self.selection = set()
@@ -313,9 +319,9 @@ def update_selection(self):
                 width = 4
                 val_x, val_y = area.value_pair
                 filts.append(
-                    filter.Values([
-                        filter.FilterDiscrete(self.attr_x.name, [val_x]),
-                        filter.FilterDiscrete(self.attr_y.name, [val_y])
+                    data_filter.Values([
+                        data_filter.FilterDiscrete(self.attr_x.name, [val_x]),
+                        data_filter.FilterDiscrete(self.attr_y.name, [val_y])
                     ]))
             else:
                 width = 1
@@ -325,7 +331,7 @@ def update_selection(self):
         if len(filts) == 1:
             filts = filts[0]
         else:
-            filts = filter.Values(filts, conjunction=False)
+            filts = data_filter.Values(filts, conjunction=False)
         selection = filts(self.discrete_data)
         idset = set(selection.ids)
         sel_idx = [i for i, id in enumerate(self.data.ids) if id in idset]
@@ -516,6 +522,23 @@ def _oper(attr, txt):
                   0, bottom)
         # Assume similar height for both lines
         text("N = " + fmt(chi.n), 0, bottom - xl.boundingRect().height())
+        cochran_ok = self._cochran_ok(chi)
+        self.Information.cochran(shown=cochran_ok)
+        self.Warning.cochran(shown= not cochran_ok)
+
+    def _cochran_ok(self, chi):
+        """
+        Return True if Cochran's rule is met; otherwise return False.
+        """
+        expected = np.asarray(chi.expected, dtype=float)
+        cells = int(expected.size)
+        if cells == 0:
+            return False
+        eps = 1e-12
+        num_lt1 = int((expected < 1.0 - eps).sum())
+        num_lt5 = int((expected < 5.0 - eps).sum())
+        ok = (num_lt1 == 0) and (num_lt5 <= 0.2 * cells)
+        return ok
 
     def get_widget_name_extension(self):
         if self.data is not None:

Orange/widgets/visualize/tests/test_owsieve.py

@@ -110,6 +110,30 @@ def test_chisquare(self):
         chi = ChiSqStats(table, 0, 1)
         self.assertFalse(isnan(chi.chisq))
 
+    def test_cochran_messages(self):
+        a = DiscreteVariable("A", values=("a1","a2","a3"))
+        b = DiscreteVariable("B", values=("b1","b2","b3"))
+
+        # PASS: 20/20/20 × 20/20/20
+        rows_ok = ["a1"]*20 + ["a2"]*20 + ["a3"]*20
+        cols_ok = ["b1"]*20 + ["b2"]*20 + ["b3"]*20
+        table_ok = Table.from_list(Domain([a,b]), list(zip(rows_ok, cols_ok)))
+        self.send_signal(self.widget.Inputs.data, table_ok)
+        self.widget.attr_x, self.widget.attr_y = a, b
+        self.widget.update_graph()
+        assert self.widget.Information.cochran.is_shown()
+        assert not self.widget.Warning.cochran.is_shown()
+
+        # FAIL: 10/20/30 × 20/20/20
+        rows_bad = ["a1"]*10 + ["a2"]*20 + ["a3"]*30
+        cols_bad = ["b1"]*20 + ["b2"]*20 + ["b3"]*20
+        table_bad = Table.from_list(Domain([a,b]), list(zip(rows_bad, cols_bad)))
+        self.send_signal(self.widget.Inputs.data, table_bad)
+        self.widget.attr_x, self.widget.attr_y = a, b
+        self.widget.update_graph()
+        assert not self.widget.Information.cochran.is_shown()
+        assert self.widget.Warning.cochran.is_shown()
+
     def test_metadata(self):
         """
         Widget should interpret meta data which are continuous or discrete in
@@ -168,10 +192,23 @@ def test_vizrank_receives_manual_change(self, auto_select):
     def test_input_features(self):
         self.assertTrue(self.widget.attr_box.isEnabled())
         self.send_signal(self.widget.Inputs.data, self.iris)
-        self.send_signal(self.widget.Inputs.features,
-                         AttributeList(self.iris.domain.attributes))
+
+        # Force a known initial state different from the incoming features
+        a0, a1, a2, a3 = self.iris.domain.attributes
+        self.widget.attr_x, self.widget.attr_y = a2, a3
+
+        # Send features -> triggers set_input_features -> resolve_shown_attributes
+        feats = AttributeList([a0, a1])
+        self.send_signal(self.widget.Inputs.features, feats)
+
+        # Attributes should now follow the provided features
+        self.assertEqual((self.widget.attr_x, self.widget.attr_y), (a0, a1))
+
+        # Existing checks
         self.assertFalse(self.widget.attr_box.isEnabled())
         self.assertFalse(self.widget.vizrank_button().isEnabled())
+
+        # Remove features -> widget returns to interactive mode
         self.send_signal(self.widget.Inputs.features, None)
         self.assertTrue(self.widget.attr_box.isEnabled())
         self.assertTrue(self.widget.vizrank_button().isEnabled())