Merge pull request #239 from zalando/chi_square_more_statistics

Added observed and expected frequencies to chi-square statistics

expan/core/experiment.py

@@ -318,7 +318,7 @@ def chi_square_test_result_and_statistics(self, variant_column, weights, min_cou
         :param alpha: significance level, 0.05 by default
         :return: True(if split is consistent with the given split) or
                  False(if split is not consistent with the given split)
-        :rtype:  Boolean, float, float
+        :rtype:  Boolean, float, float, pd.Series, pd.Series
         """
         if not hasattr(variant_column, '__len__'):
             raise ValueError("Variant split check was cancelled since input variant column is empty or doesn't exist.")
@@ -351,4 +351,4 @@ def chi_square_test_result_and_statistics(self, variant_column, weights, min_cou
         # Compute chi-square and p-value statistics
         chi_square_val, p_val = statx.chi_square(observed_freqs.sort_index(), expected_freqs.sort_index())
 
-        return p_val >= alpha, p_val, chi_square_val
+        return p_val >= alpha, p_val, chi_square_val, observed_freqs, expected_freqs

tests/tests_core/test_experiment.py

@@ -368,6 +368,13 @@ def test_chi_square_test_result_and_statistics_same_weights(self):
         self.assertEqual(result[0], True)
         self.assertAlmostEqual(result[1], 0.8087921354109989)
         self.assertAlmostEqual(result[2], 1.6)
+        self.assertEqual(result[3]['A'], 23)
+        self.assertEqual(result[3]['B'], 18)
+        self.assertEqual(result[3]['C'], 17)
+        self.assertEqual(result[3]['D'], 19)
+        self.assertEqual(result[3]['E'], 23)
+
+        self.assertTrue(all([val == 20 for val in result[4]]))
 
     def test_chi_square_test_result_and_statistics_different_weights(self):
         exp = self.getExperiment()
@@ -377,6 +384,8 @@ def test_chi_square_test_result_and_statistics_different_weights(self):
         self.assertEqual(result[0], False)
         self.assertAlmostEqual(result[1], 9.064563321754584e-07)
         self.assertAlmostEqual(result[2], 33.585)
+        self.assertEqual(len(result[3]), 5)
+        self.assertEqual(len(result[4]), 5)
 
     def test_chi_square_test_result_and_statistics_2_categories(self):
         exp = self.getExperiment()
@@ -386,6 +395,10 @@ def test_chi_square_test_result_and_statistics_2_categories(self):
         self.assertEqual(result[0], True)
         self.assertAlmostEqual(result[1], 1.0)
         self.assertAlmostEqual(result[2], 0.0)
+        self.assertEqual(result[3]['A'], 17)
+        self.assertEqual(result[4]['A'], 17)
+        self.assertEqual(result[3]['B'], 17)
+        self.assertEqual(result[4]['B'], 17)
 
     def test_chi_square_test_result_and_statistics_NaN_data(self):
         exp = self.getExperiment()