Merge branch 'main' into gaugup/ReferPredictedValueCFData

microsoft · Oct 30, 2023 · 1bf94d3 · 1bf94d3
2 parents a21327a + 4d8d9b7
commit 1bf94d3
Show file tree

Hide file tree

Showing 2 changed files with 24 additions and 2 deletions.
diff --git a/responsibleai/requirements.txt b/responsibleai/requirements.txt
@@ -1,4 +1,4 @@
-dice-ml>=0.10,<0.11
+dice-ml>=0.11,<0.12
 econml>=0.14.1
 statsmodels<0.14.0
 jsonschema

diff --git a/responsibleai/tests/counterfactual/test_counterfactual_advanced_features.py b/responsibleai/tests/counterfactual/test_counterfactual_advanced_features.py
@@ -21,10 +21,15 @@ class TestCounterfactualAdvancedFeatures(object):
 
     @pytest.mark.parametrize('vary_all_features', [True, False])
     @pytest.mark.parametrize('feature_importance', [True, False])
+    @pytest.mark.parametrize('encode_target_as_strings', [True, False])
     def test_counterfactual_vary_features(
-            self, vary_all_features, feature_importance):
+            self, vary_all_features, feature_importance,
+            encode_target_as_strings):
         X_train, X_test, y_train, y_test, feature_names, _ = \
             create_iris_data()
+        if encode_target_as_strings:
+            y_train = y_train.astype(str)
+            y_test = y_test.astype(str)
 
         model = create_lightgbm_classifier(X_train, y_train)
         X_train['target'] = y_train
@@ -50,6 +55,23 @@ def test_counterfactual_vary_features(
 
         cf_obj = rai_insights.counterfactual.get()[0]
         assert cf_obj is not None
+        for index in range(0, len(cf_obj.cf_examples_list)):
+            assert isinstance(
+                cf_obj.cf_examples_list[
+                    index].test_instance_df[
+                        'target'].values[0], str) == encode_target_as_strings
+
+            assert cf_obj.cf_examples_list[
+                index].test_instance_df['target'].values[0] in set(
+                    y_train)
+
+            cf_target_array = cf_obj.cf_examples_list[0].final_cfs_df[
+                'target'].values
+            for inner_index in range(0, 10):
+                assert isinstance(
+                    cf_target_array[
+                        inner_index], str) == encode_target_as_strings
+                assert cf_target_array[inner_index] in set(y_train)
 
     @pytest.mark.parametrize('feature_importance', [True, False])
     def test_counterfactual_permitted_range(self, feature_importance):