Merge pull request #9 from BBVA/fix/linear_combinations_show_continou…

…us_lin_comb

Show in LinearCombinationsTest the continuous variables with linear combinations

mercury/robust/data_tests.py

@@ -134,7 +134,14 @@ def run(self, *args, **kwargs):
             )
 
         if lin_combinations is not None:
-            raise FailedTestError("Test failed. Linear combinations for continuous features were encountered.")
+            # Create message with the linear combinations found
+            lin_combinations = np.around(lin_combinations, decimals=5)
+            lin_combs_found = []
+            for i in range(lin_combinations.shape[0]):
+                lin_comb_idx = np.where(lin_combinations[i] != 0)[0]
+                lin_comb_cols = self.base_dataset.loc[:, numeric_feats].columns[lin_comb_idx].tolist()
+                lin_combs_found.append(lin_comb_cols)
+            raise FailedTestError(f"Test failed. Linear combinations for continuous features were encountered: {lin_combs_found}.")
 
         individually_redundant = CategoryStruct.individually_redundant(self.base_dataset, current_schema.categorical_feats)
         if len(individually_redundant) > 0:

tests/test_data_tests.py

@@ -857,4 +857,20 @@ def test_no_duplicated_test():
     with pytest.raises(FailedTestError):
         test.run()
 
+def test_lin_combinations_cont():
 
+    df = pd.DataFrame()
+    df["f1"] = np.random.uniform(size=100)
+    df["f2"] = df["f1"] * 2
+    df["f3"] = np.random.uniform(size=100)
+    df["f4"] = df["f1"] + df["f2"]
+    df["f5"] = np.random.uniform(size=100)
+
+    df["f6"] = np.random.uniform(size=100)
+    df["f7"] = df["f6"] * 3
+    df["f8"] = df["f6"] * 0.1
+
+    schma_reference = DataSchema().generate(df).calculate_statistics()
+    linear_comb_test = LinearCombinationsTest(df, dataset_schema=schma_reference)
+    with pytest.raises(FailedTestError, match="'f1', 'f2'"):
+        linear_comb_test.run()