Fix the dimensions mismatch bug for dataset with categorical columns (#…

…274)

* updated the dimensions bug

* fixed flake error

* added tests

* moved functions to conftest

Co-authored-by: Gaurav Gupta <[email protected]>

dice_ml/data_interfaces/public_data_interface.py

@@ -49,21 +49,10 @@ def __init__(self, params):
             name) for name in self.categorical_feature_names if name in self.data_df]
 
         self._validate_and_set_continuous_features_precision(params=params)
-
-        if len(self.categorical_feature_names) > 0:
-            for feature in self.categorical_feature_names:
-                self.data_df[feature] = self.data_df[feature].apply(str)
-            self.data_df[self.categorical_feature_names] = self.data_df[self.categorical_feature_names].astype(
-                'category')
-
-        if len(self.continuous_feature_names) > 0:
-            for feature in self.continuous_feature_names:
-                if self.get_data_type(feature) == 'float':
-                    self.data_df[feature] = self.data_df[feature].astype(
-                        np.float32)
-                else:
-                    self.data_df[feature] = self.data_df[feature].astype(
-                        np.int32)
+        self.data_df = self._set_feature_dtypes(
+                self.data_df,
+                self.categorical_feature_names,
+                self.continuous_feature_names)
 
         # should move the below snippet to gradient based dice interfaces
         # self.one_hot_encoded_data = self.one_hot_encode_data(self.data_df)
@@ -149,6 +138,25 @@ def _validate_and_set_permitted_range(self, params):
                     )
         self.permitted_range, _ = self.get_features_range(input_permitted_range)
 
+    def _set_feature_dtypes(self, data_df, categorical_feature_names,
+                            continuous_feature_names):
+        """Set the correct type of each feature column."""
+        if len(categorical_feature_names) > 0:
+            for feature in categorical_feature_names:
+                data_df[feature] = data_df[feature].apply(str)
+            data_df[categorical_feature_names] = data_df[categorical_feature_names].astype(
+                'category')
+
+        if len(continuous_feature_names) > 0:
+            for feature in continuous_feature_names:
+                if self.get_data_type(feature) == 'float':
+                    data_df[feature] = data_df[feature].astype(
+                        np.float32)
+                else:
+                    data_df[feature] = data_df[feature].astype(
+                        np.int32)
+        return data_df
+
     def check_features_to_vary(self, features_to_vary):
         if features_to_vary is not None and features_to_vary != 'all':
             not_training_features = set(features_to_vary) - set(self.feature_names)
@@ -546,6 +554,10 @@ def prepare_query_instance(self, query_instance):
             raise ValueError("Query instance should be a dict, a pandas dataframe, a list, or a list of dicts")
 
         test = test.reset_index(drop=True)
+        # encode categorical and numerical columns
+        test = self._set_feature_dtypes(test,
+                                        self.categorical_feature_names,
+                                        self.continuous_feature_names)
         return test
 
         # TODO: create a new method, get_LE_min_max_normalized_data() to get label-encoded and normalized data. Keep this

dice_ml/explainer_interfaces/dice_KD.py

@@ -82,7 +82,10 @@ def _generate_counterfactuals(self, query_instance, total_CFs, desired_range=Non
 
         # Prepares user defined query_instance for DiCE.
         query_instance_orig = query_instance.copy()
-        query_instance = self.data_interface.prepare_query_instance(query_instance=query_instance)
+        query_instance_orig = self.data_interface.prepare_query_instance(
+                query_instance=query_instance_orig)
+        query_instance = self.data_interface.prepare_query_instance(
+                query_instance=query_instance)
 
         # find the predicted value of query_instance
         test_pred = self.predict_fn(query_instance)[0]
@@ -103,7 +106,6 @@ def _generate_counterfactuals(self, query_instance, total_CFs, desired_range=Non
         # Partitioned dataset and KD Tree for each class (binary) of the dataset
         self.dataset_with_predictions, self.KD_tree, self.predictions = \
             self.build_KD_tree(data_df_copy, desired_range, desired_class, self.predicted_outcome_name)
-
         query_instance, cfs_preds = self.find_counterfactuals(data_df_copy,
                                                               query_instance, query_instance_orig,
                                                               desired_range,
@@ -224,7 +226,6 @@ def find_counterfactuals(self, data_df_copy, query_instance, query_instance_orig
         for col in pd.get_dummies(data_df_copy[self.data_interface.feature_names]).columns:
             if col not in query_instance_df_dummies.columns:
                 query_instance_df_dummies[col] = 0
-
         self.final_cfs, cfs_preds = self.vary_valid(query_instance_df_dummies,
                                                     total_CFs,
                                                     features_to_vary,

dice_ml/explainer_interfaces/dice_genetic.py

@@ -268,7 +268,10 @@ def _generate_counterfactuals(self, query_instance, total_CFs, initialization="k
 
         # Prepares user defined query_instance for DiCE.
         query_instance_orig = query_instance
-        query_instance = self.data_interface.prepare_query_instance(query_instance=query_instance)
+        query_instance_orig = self.data_interface.prepare_query_instance(
+                query_instance=query_instance_orig)
+        query_instance = self.data_interface.prepare_query_instance(
+                query_instance=query_instance)
         query_instance = self.label_encode(query_instance)
         query_instance = np.array(query_instance.values[0])
         self.x1 = query_instance

dice_ml/utils/helpers.py

@@ -7,7 +7,6 @@
 import numpy as np
 import pandas as pd
 from sklearn.model_selection import train_test_split
-# for data transformations
 from sklearn.preprocessing import FunctionTransformer
 
 import dice_ml
@@ -129,6 +128,13 @@ def get_custom_dataset_modelpath_pipeline():
     return modelpath
 
 
+def get_custom_vars_dataset_modelpath_pipeline():
+    pkg_path = dice_ml.__path__[0]
+    model_ext = '.sav'
+    modelpath = os.path.join(pkg_path, 'utils', 'sample_trained_models', 'custom_vars'+model_ext)
+    return modelpath
+
+
 def get_custom_dataset_modelpath_pipeline_binary():
     pkg_path = dice_ml.__path__[0]
     model_ext = '.sav'
@@ -168,7 +174,6 @@ def get_base_gen_cf_initialization(data_interface, encoded_size, cont_minx, cont
                                    wm1, wm2, wm3, learning_rate):
     # Dice Imports - TODO: keep this method for VAE as a spearate module or move it to feasible_base_vae.py.
     #                      Check dependencies.
-    # Pytorch
     from torch import optim
 
     from dice_ml.utils.sample_architecture.vae_model import CF_VAE

tests/conftest.py

@@ -1,9 +1,15 @@
+import pickle
 from collections import OrderedDict
 
 import pandas as pd
 import pytest
+from sklearn.compose import ColumnTransformer
 from sklearn.datasets import fetch_california_housing, load_iris
+from sklearn.ensemble import RandomForestClassifier
+from sklearn.impute import SimpleImputer
 from sklearn.model_selection import train_test_split
+from sklearn.pipeline import Pipeline
+from sklearn.preprocessing import OneHotEncoder, StandardScaler
 
 import dice_ml
 from dice_ml.utils import helpers
@@ -110,6 +116,33 @@ def private_data_object():
     return dice_ml.Data(features=features_dict, outcome_name='income')
 
 
+@pytest.fixture()
+def load_custom_vars_testing_dataset():
+    data = [['a', 0, 10, 0], ['b', 1, 10000, 0], ['c', 0, 14, 0], ['a', 2, 88, 0], ['c', 1, 14, 0]]
+    return pd.DataFrame(data, columns=['Categorical', 'CategoricalNum', 'Numerical', 'Outcome'])
+
+
+@pytest.fixture()
+def _save_custom_vars_dataset_model():
+    numeric_trans = Pipeline(steps=[('imputer', SimpleImputer(strategy='median')),
+                                    ('scaler', StandardScaler())])
+    cat_trans = Pipeline(steps=[('imputer',
+                                SimpleImputer(fill_value='missing',
+                                              strategy='constant')),
+                                ('onehot', OneHotEncoder(handle_unknown='ignore'))])
+    transformations = ColumnTransformer(transformers=[('num', numeric_trans,
+                                                      ['Numerical']),
+                                                      ('cat', cat_trans,
+                                                      pd.Index(['Categorical', 'CategoricalNum'], dtype='object'))])
+    clf = Pipeline(steps=[('preprocessor', transformations),
+                          ('regressor', RandomForestClassifier())])
+    dataset = load_custom_vars_testing_dataset()
+    model = clf.fit(dataset[["Categorical", "CategoricalNum", "Numerical"]],
+                    dataset["Outcome"])
+    modelpath = helpers.get_custom_vars_dataset_modelpath_pipeline()
+    pickle.dump(model, open(modelpath, 'wb'))
+
+
 @pytest.fixture()
 def sample_adultincome_query():
     """
@@ -188,6 +221,14 @@ def sample_custom_query_10():
     )
 
 
+@pytest.fixture()
+def sample_custom_vars_query_1():
+    """
+    Returns a sample query instance for the custom dataset
+    """
+    return pd.DataFrame({'Categorical': ['a'], 'CategoricalNum': [0], 'Numerical': [25]})
+
+
 @pytest.fixture()
 def sample_counterfactual_example_dummy():
     """

tests/test_dice_interface/test_dice_KD.py

@@ -18,6 +18,17 @@ def KD_binary_classification_exp_object():
     return exp
 
 
+@pytest.fixture()
+def KD_binary_vars_classification_exp_object(load_custom_vars_testing_dataset):
+    backend = 'sklearn'
+    dataset = load_custom_vars_testing_dataset
+    d = dice_ml.Data(dataframe=dataset, continuous_features=['Numerical'], outcome_name='Outcome')
+    ML_modelpath = helpers.get_custom_vars_dataset_modelpath_pipeline()
+    m = dice_ml.Model(model_path=ML_modelpath, backend=backend)
+    exp = dice_ml.Dice(d, m, method='kdtree')
+    return exp
+
+
 @pytest.fixture()
 def KD_multi_classification_exp_object():
     backend = 'sklearn'
@@ -194,3 +205,44 @@ def test_KD_tree_counterfactual_explanations_output(self, desired_range, sample_
     def test_zero_cfs(self, desired_class, desired_range, sample_custom_query_4, total_CFs):
         self.exp_regr._generate_counterfactuals(query_instance=sample_custom_query_4, total_CFs=total_CFs,
                                                 desired_range=desired_range)
+
+
+class TestDiceKDBinaryVarsClassificationMethods:
+    @pytest.fixture(autouse=True)
+    def _initiate_exp_object(self, KD_binary_vars_classification_exp_object):
+        self.exp = KD_binary_vars_classification_exp_object  # explainer object
+        self.data_df_copy = self.exp.data_interface.data_df.copy()
+
+    # When a query's feature value is not within the permitted range and the feature is not allowed to vary
+    @pytest.mark.parametrize(("desired_range", "desired_class", "total_CFs", "features_to_vary", "permitted_range"),
+                             [(None, 0, 4, ['Numerical'], {'CategoricalNum': ['1', '2']})])
+    def test_invalid_query_instance(self, desired_range, desired_class, sample_custom_vars_query_1, total_CFs,
+                                    features_to_vary, permitted_range):
+        self.exp.dataset_with_predictions, self.exp.KD_tree, self.exp.predictions = \
+            self.exp.build_KD_tree(self.data_df_copy, desired_range, desired_class, self.exp.predicted_outcome_name)
+
+        with pytest.raises(ValueError, match="is outside the permitted range and isn't allowed to vary"):
+            self.exp._generate_counterfactuals(query_instance=sample_custom_vars_query_1, total_CFs=total_CFs,
+                                               features_to_vary=features_to_vary, permitted_range=permitted_range)
+
+    # Verifying the output of the KD tree
+    @pytest.mark.parametrize(("desired_class", "total_CFs"), [(0, 1)])
+    @pytest.mark.parametrize('posthoc_sparsity_algorithm', ['linear', 'binary', None])
+    def test_KD_tree_output(self, desired_class, sample_custom_vars_query_1, total_CFs, posthoc_sparsity_algorithm):
+        self.exp._generate_counterfactuals(query_instance=sample_custom_vars_query_1, desired_class=desired_class,
+                                           total_CFs=total_CFs,
+                                           posthoc_sparsity_algorithm=posthoc_sparsity_algorithm)
+        self.exp.final_cfs_df.Numerical = self.exp.final_cfs_df.Numerical.astype(int)
+        expected_output = self.exp.data_interface.data_df
+
+        assert all(self.exp.final_cfs_df.Numerical == expected_output.Numerical[0])
+        assert all(self.exp.final_cfs_df.Categorical == expected_output.Categorical[0])
+
+    # Verifying the output of the KD tree
+    @pytest.mark.parametrize(("desired_class", "total_CFs"), [(0, 1)])
+    def test_KD_tree_counterfactual_explanations_output(self, desired_class, sample_custom_vars_query_1, total_CFs):
+        counterfactual_explanations = self.exp.generate_counterfactuals(
+            query_instances=sample_custom_vars_query_1, desired_class=desired_class,
+            total_CFs=total_CFs)
+
+        assert counterfactual_explanations is not None