Apply transformation first

Author: hamidkm9

src/linearboost/linear_boost.py

@@ -271,38 +271,36 @@ def _check_X_y(self, X, y) -> tuple[np.ndarray, np.ndarray]:
     def fit(self, X, y, sample_weight=None) -> Self:
         if self.algorithm not in {"SAMME", "SAMME.R"}:
             raise ValueError("algorithm must be 'SAMME' or 'SAMME.R'")
-        X, y = check_X_y(X, y, accept_sparse=True)
+
+        if self.scaler not in _scalers:
+            raise ValueError('Invalid scaler provided; got "%s".' % self.scaler)
+
+        if self.scaler == "minmax":
+            self.scaler_ = clone(_scalers["minmax"])
+        else:
+            self.scaler_ = make_pipeline(
+                clone(_scalers[self.scaler]), clone(_scalers["minmax"])
+            )
+        X_transformed = self.scaler_.fit_transform(X)
 
         if sample_weight is not None:
             sample_weight = np.asarray(sample_weight)
-            if sample_weight.shape[0] != X.shape[0]:
+            if sample_weight.shape[0] != X_transformed.shape[0]:
                 raise ValueError(
-                    f"sample_weight.shape == {sample_weight.shape} is incompatible with X.shape == {X.shape}"
+                    f"sample_weight.shape == {sample_weight.shape} is incompatible with X.shape == {X_transformed.shape}"
                 )
-            # fix here
             nonzero_mask = (
                 sample_weight.sum(axis=1) != 0
                 if sample_weight.ndim > 1
                 else sample_weight != 0
             )
-            X = X[nonzero_mask]
+            X_transformed = X_transformed[nonzero_mask]
             y = y[nonzero_mask]
             sample_weight = sample_weight[nonzero_mask]
-        X, y = self._check_X_y(X, y)
+        X_transformed, y = self._check_X_y(X_transformed, y)
         self.classes_ = np.unique(y)
         self.n_classes_ = self.classes_.shape[0]
 
-        if self.scaler not in _scalers:
-            raise ValueError('Invalid scaler provided; got "%s".' % self.scaler)
-
-        if self.scaler == "minmax":
-            self.scaler_ = clone(_scalers["minmax"])
-        else:
-            self.scaler_ = make_pipeline(
-                clone(_scalers[self.scaler]), clone(_scalers["minmax"])
-            )
-        X_transformed = self.scaler_.fit_transform(X)
-
         if self.class_weight is not None:
             valid_presets = ("balanced", "balanced_subsample")
             if (
@@ -326,7 +324,7 @@ def fit(self, X, y, sample_weight=None) -> Self:
                 warnings.filterwarnings(
                     "ignore",
                     category=FutureWarning,
-                    message=".*parameter 'algorithm' is deprecated.*",
+                    message=".*parameter 'algorithm' may change in the future.*",
                 )
             return super().fit(X_transformed, y, sample_weight)