ignore min_category_size_ratio for balance_classes=True (#2336)

deepchecks/tabular/checks/model_evaluation/train_test_prediction_drift.py

@@ -73,11 +73,10 @@ class TrainTestPredictionDrift(TrainTestCheck, ReduceMixin):
         disproportionally. This filter is applied to both distributions, in both margins.
     min_category_size_ratio: float, default 0.01
         minimum size ratio for categories. Categories with size ratio lower than this number are binned
-        into an "Other" category.
+        into an "Other" category. Ignored if balance_classes=True.
     max_num_categories_for_drift: int, default: None
         Only relevant if drift is calculated for classification predictions. Max number of allowed categories.
-        If there are more,
-        they are binned into an "Other" category. This limit applies for both drift calculation and distribution plots.
+        If there are more, they are binned into an "Other" category.
     max_num_categories_for_display: int, default: 10
         Max number of categories to show in plot.
     show_categories_by: str, default: 'largest_difference'

deepchecks/tabular/checks/train_test_validation/train_test_label_drift.py

@@ -57,10 +57,10 @@ class TrainTestLabelDrift(TrainTestCheck, ReduceLabelMixin):
         disproportionally. This filter is applied to both distributions, in both margins.
     min_category_size_ratio: float, default 0.01
         minimum size ratio for categories. Categories with size ratio lower than this number are binned
-        into an "Other" category.
+        into an "Other" category. Ignored if balance_classes=True.
     max_num_categories_for_drift: int, default: None
         Only for classification. Max number of allowed categories. If there are more,
-        they are binned into an "Other" category. This limit applies for both drift calculation and distribution plots
+        they are binned into an "Other" category.
     max_num_categories_for_display: int, default: 10
         Max number of categories to show in plot.
     show_categories_by: str, default: 'largest_difference'

deepchecks/utils/distribution/drift.py

@@ -129,6 +129,9 @@ def cramers_v(dist1: Union[np.ndarray, pd.Series], dist2: Union[np.ndarray, pd.S
         the bias-corrected Cramer's V value of the 2 distributions.
 
     """
+    # If balance_classes is True, min_category_size_ratio should not affect results:
+    min_category_size_ratio = min_category_size_ratio if balance_classes is False else 0
+
     dist1_counts, dist2_counts, _ = preprocess_2_cat_cols_to_same_bins(dist1, dist2, min_category_size_ratio,
                                                                        max_num_categories, sort_by)
 

deepchecks/vision/checks/model_evaluation/train_test_prediction_drift.py

@@ -92,16 +92,15 @@ class TrainTestPredictionDrift(TrainTestCheck, ReducePropertyMixin):
         disproportionally. This filter is applied to both distributions, in both margins.
     min_category_size_ratio: float, default 0.01
         minimum size ratio for categories. Categories with size ratio lower than this number are binned
-        into an "Other" category.
+        into an "Other" category. Ignored if balance_classes=True.
     max_num_categories_for_drift: int, default: None
         Only for discrete properties. Max number of allowed categories. If there are more,
-        they are binned into an "Other" category. This limit applies for both drift calculation and distribution plots.
+        they are binned into an "Other" category.
     max_num_categories_for_display: int, default: 10
         Max number of categories to show in plot.
     show_categories_by: str, default: 'largest_difference'
         Specify which categories to show for categorical features' graphs, as the number of shown categories is limited
         by max_num_categories_for_display. Possible values:
-
         - 'train_largest': Show the largest train categories.
         - 'test_largest': Show the largest test categories.
         - 'largest_difference': Show the largest difference between categories.

deepchecks/vision/checks/train_test_validation/train_test_label_drift.py

@@ -86,10 +86,10 @@ class TrainTestLabelDrift(TrainTestCheck, ReducePropertyMixin, ReduceLabelMixin)
         disproportionally. This filter is applied to both distributions, in both margins.
     min_category_size_ratio : float, default 0.01
         minimum size ratio for categories. Categories with size ratio lower than this number are binned
-        into an "Other" category.
+        into an "Other" category. Ignored if balance_classes=True.
     max_num_categories_for_drift : int, default: None
         Only for discrete properties. Max number of allowed categories. If there are more,
-        they are binned into an "Other" category. This limit applies for both drift calculation and distribution plots.
+        they are binned into an "Other" category.
     max_num_categories_for_display: int, default: 10
         Max number of categories to show in plot.
     show_categories_by : str, default: 'largest_difference'

tests/utils/drift_test.py

@@ -91,6 +91,18 @@ def test_cramers_v_min_category_ratio():
     res_min_cat_ratio = cramers_v(dist1=dist1, dist2=dist2, min_category_size_ratio=0.1)
     assert_that(res_min_cat_ratio, close_to(0.208, 0.01))
 
+def test_cramers_v_imbalanced():
+    dist1 = np.array([0] * 9900 + [1] * 100)
+    dist2 = np.array([0] * 9950 + [1] * 50)
+    res = cramers_v(dist1=dist1, dist2=dist2, balance_classes=True)
+    assert_that(res, close_to(0.17, 0.01))
+
+def test_cramers_v_imbalanced_ignore_min_category_size():
+    dist1 = np.array([0] * 9900 + [1] * 100)
+    dist2 = np.array([0] * 9950 + [1] * 50)
+    res = cramers_v(dist1=dist1, dist2=dist2, balance_classes=True, min_category_size_ratio=0.1)
+    assert_that(res, close_to(0.17, 0.01))
+
 
 def test_ks_no_drift():
     dist1 = np.zeros(100)