[FIX] Fix Chi2 computation for variables with values with no instances

Orange/widgets/visualize/owsieve.py

@@ -42,6 +42,7 @@ def __init__(self, data, attr1, attr2):
         self.expected = np.outer(self.probs_y, self.probs_x) * self.n
         self.residuals = \
             (self.observed - self.expected) / np.sqrt(self.expected)
+        self.residuals = np.nan_to_num(self.residuals)
         self.chisqs = self.residuals ** 2
         self.chisq = float(np.sum(self.chisqs))
         self.p = chi2.sf(

Orange/widgets/visualize/tests/test_owsieve.py

@@ -1,5 +1,6 @@
 # Test methods with long descriptive names can omit docstrings
 # pylint: disable=missing-docstring
+from math import isnan
 import numpy as np
 
 from AnyQt.QtCore import QEvent, QPoint, Qt
@@ -8,6 +9,7 @@
 from Orange.data import DiscreteVariable, Domain, Table
 from Orange.widgets.tests.base import WidgetTest, WidgetOutputsTestMixin
 from Orange.widgets.visualize.owsieve import OWSieveDiagram
+from Orange.widgets.visualize.owsieve import ChiSqStats
 
 
 class TestOWSieveDiagram(WidgetTest, WidgetOutputsTestMixin):
@@ -46,3 +48,14 @@ def test_keyerror(self):
         data = Table("iris")
         data = data[0:1]
         self.send_signal("Data", data)
+
+    def test_chisquare(self):
+        """
+         gh-2031
+         Check if it can calculate chi square when there are no attributes which suppose to be.
+        """
+        a = DiscreteVariable("a", values=["y", "n"])
+        b = DiscreteVariable("b", values=["y", "n", "o"])
+        table = Table(Domain([a, b]), list(zip("yynny", "ynyyn")))
+        chi = ChiSqStats(table, 0, 1)
+        self.assertFalse(isnan(chi.chisq))