data validation issue patched

Author: Julien Roussel

qolmat/benchmark/missing_patterns.py

@@ -247,14 +247,15 @@ def generate_mask(self, X: pd.DataFrame) -> pd.DataFrame:
             ratio_masked = self.ratio_masked
             if self.sample_proportional:
                 ratio_masked *= self.dict_ratios[col] * len(X.columns)
-            n_masked_col = math.ceil(self.ratio_masked * len(X))
-            indices = np.where(X[col].notna())[0]
-            indices = self.random_state.choice(
-                indices,
+            n_masked_col = math.ceil(ratio_masked * len(X))
+            indices_int = np.where(X[col].notna())[0]
+            indices_int = self.random_state.choice(
+                indices_int,
                 replace=False,
                 size=n_masked_col,
             )
-            df_mask[col].iloc[indices] = True
+            indices_int = df_mask.index[indices_int]
+            df_mask.loc[indices_int, col] = True
 
         return df_mask
 

qolmat/imputations/preprocessing.py

@@ -167,7 +167,7 @@ def fit(self, X: NDArray, y: Optional[NDArray] = None) -> "BinTransformer":
             X,
             accept_sparse=False,
             ensure_all_finite="allow-nan",
-            reset=False,
+            reset=True,
             dtype=["float", "int", "string", "categorical", "object"],
         )
         df = utils._validate_input(X)

qolmat/utils/data.py

@@ -198,9 +198,9 @@ def get_data(
         df = pd.read_csv(path)
         df = df[
             ["Survived", "Sex", "Age", "SibSp", "Parch", "Fare", "Embarked"]
-        ]
+        ].copy()
         df["Age"] = pd.to_numeric(df["Age"], errors="coerce")
-        df["Fare"] = pd.to_numeric(df["Fare"], errors="coerce")
+        df.loc["Fare"] = pd.to_numeric(df["Fare"], errors="coerce")
         return df
     elif name_data == "Artificial":
         city = "Wonderland"

tests/imputations/test_imputers_diffusions.py

@@ -26,6 +26,7 @@
         "col5": [93, 75, np.nan, 12, np.nan],
     },
     index=pd.date_range("2023-04-17", periods=5, freq="D"),
+    dtype=float,
 )
 df_incomplete.index = df_incomplete.index.set_names("datetime")
 

tests/utils/test_data.py

@@ -336,7 +336,6 @@ def test_data_get_data(
             "Beijing_online",
             "Superconductor_online",
             "Monach_weather",
-            "Monach_weather",
             "Monach_electricity_australia",
             "Titanic",
             "SNCF",
@@ -372,7 +371,7 @@ def test_data_get_data(
         pd.testing.assert_frame_equal(df_result, df_monach_elec_preprocess)
     elif name_data == "Titanic":
         assert mock_read_dl.call_count == 1
-        assert np.shape(df_result) == (3, 7)
+        assert np.shape(df_result) == (4, 7)
     elif name_data == "SNCF":
         assert not df_result.empty
         assert df_result.index.name == "station"