update from base ALE

Author: abaziree

src/hidimstat/visualization/accumulated_local_effects.py

@@ -168,7 +168,7 @@ def compute_ale_1d(
     confidence_interval=True,
     confidence_level=0.95,
     n_bootstraps=20,
-    n_jobs=-1
+    n_jobs=-1,
 ):
     """Compute the 1D Accumulated Local Effect for a single numerical feature.
 
@@ -304,7 +304,9 @@ def _compute_ale_curve(X_sample):
             mask_low = x_sample != grid_values[0]
             mask_high = x_sample != grid_values[-1]
             X_low[mask_low, feature_idx] = grid_values[value_idx[mask_low] - 1]
-            X_high[mask_high, feature_idx] = grid_values[value_idx[mask_high] + 1]
+            X_high[mask_high, feature_idx] = grid_values[
+                value_idx[mask_high] + 1
+            ]
 
             local_effects_low = _predict_fn(
                 estimator, X_sample[mask_low]
@@ -340,7 +342,10 @@ def _compute_ale_curve(X_sample):
 
         # Center: subtract the sample-weighted mean
         ale_centers = (ale_curve[1:] + ale_curve[:-1]) / 2
-        ale_curve -= np.sum(ale_centers * bin_counts[min_weight_bin:]) / bin_counts.sum()
+        ale_curve -= (
+            np.sum(ale_centers * bin_counts[min_weight_bin:])
+            / bin_counts.sum()
+        )
 
         return ale_curve
 

test/visualization/test_accumulated_local_effects.py

@@ -338,7 +338,12 @@ def test_compute_ale_1d_discrete(ale_test_data):
     # To test _bootstrap()
     with parallel_backend("threading"):
         result_bootstrap = compute_ale_1d(
-            model, X, feature_idx=0, feature_type="categorical", confidence_interval=True, n_bootstraps=2
+            model,
+            X,
+            feature_idx=0,
+            feature_type="categorical",
+            confidence_interval=True,
+            n_bootstraps=2,
         )
 
 
@@ -351,10 +356,17 @@ def test_compute_ale_1d_discrete_error(ale_test_data):
     X_const[:, 0] = 7
 
     with pytest.raises(ValueError, match="has fewer than 2 unique values"):
-        compute_ale_1d(model, X_const, feature_idx=0, feature_type="categorical")
+        compute_ale_1d(
+            model, X_const, feature_idx=0, feature_type="categorical"
+        )
     with pytest.raises(ValueError, match="must be strictly greater than"):
         compute_ale_1d(
-            model, X, feature_idx=0, feature_type="categorical", confidence_interval=True, n_bootstraps=0
+            model,
+            X,
+            feature_idx=0,
+            feature_type="categorical",
+            confidence_interval=True,
+            n_bootstraps=0,
         )
 
     with pytest.raises(