Merge branch 'main' into pull-request/6211

Author: csadorf

python/cuml/cuml/dask/neighbors/kneighbors_classifier.py

@@ -96,10 +96,8 @@ def fit(self, X, y):
                     uniq_labels.append(_y.iloc[:, i].unique())
 
         uniq_labels = da.compute(uniq_labels)[0]
-        if hasattr(uniq_labels[0], "values_host"):  # for cuDF Series
-            uniq_labels = list(map(lambda x: x.values_host, uniq_labels))
-        elif hasattr(uniq_labels[0], "values"):  # for pandas Series
-            uniq_labels = list(map(lambda x: x.values, uniq_labels))
+        if hasattr(uniq_labels[0], "to_numpy"):  # for pandas and cuDF Series
+            uniq_labels = list(map(lambda x: x.to_numpy(), uniq_labels))
         elif isinstance(uniq_labels[0], cp.ndarray):  # for CuPy arrays
             uniq_labels = list(map(lambda x: cp.asnumpy(x), uniq_labels))
         self.uniq_labels = np.sort(np.array(uniq_labels))

python/cuml/cuml/preprocessing/LabelEncoder.py

@@ -268,7 +268,7 @@ def inverse_transform(self, y: "cudf.Series"):
         category_num = len(self.classes_)
         if self.handle_unknown == "error":
             if not isinstance(ord_label, (cp.ndarray, np.ndarray)):
-                ord_label = ord_label.values_host
+                ord_label = ord_label.to_numpy()
             for ordi in ord_label:
                 if ordi < 0 or ordi >= category_num:
                     raise ValueError(

python/cuml/cuml/preprocessing/TargetEncoder.py

@@ -530,7 +530,7 @@ def _rename_col(df, col):
         res = []
         unq_vals = train[self.fold_col].unique()
         if not isinstance(unq_vals, (cp.ndarray, np.ndarray)):
-            unq_vals = unq_vals.values_host
+            unq_vals = unq_vals.to_numpy()
         for f in unq_vals:
             mask = train[self.fold_col].values == f
             dg = train.loc[~mask].groupby(x_cols).agg({self.y_col: self.stat})

python/cuml/cuml/preprocessing/encoders.py

@@ -536,7 +536,7 @@ def get_feature_names(self, input_features=None):
         feature_names = []
         for i in range(len(cats)):
             names = [
-                input_features[i] + "_" + str(t) for t in cats[i].values_host
+                input_features[i] + "_" + str(t) for t in cats[i].to_numpy()
             ]
             if self.drop_idx_ is not None and self.drop_idx_[i] is not None:
                 names.pop(self.drop_idx_[i])

python/cuml/cuml/testing/utils.py

@@ -473,7 +473,7 @@ def from_df_to_numpy(df):
     if isinstance(df, pd.DataFrame):
         return list(zip(*[df[feature] for feature in df.columns]))
     else:
-        return list(zip(*[df[feature].values_host for feature in df.columns]))
+        return list(zip(*[df[feature].to_numpy() for feature in df.columns]))
 
 
 def compare_svm(

python/cuml/tests/test_one_hot_encoder.py

@@ -1,4 +1,4 @@
-# SPDX-FileCopyrightText: Copyright (c) 2020-2025, NVIDIA CORPORATION.
+# SPDX-FileCopyrightText: Copyright (c) 2020-2026, NVIDIA CORPORATION.
 # SPDX-License-Identifier: Apache-2.0
 import math
 
@@ -26,7 +26,7 @@ def _from_df_to_cupy(df):
                 df[col] = [ord(c) for c in df[col]]
             else:
                 df[col] = [
-                    ord(c) if c is not None else c for c in df[col].values_host
+                    ord(c) if c is not None else c for c in df[col].to_numpy()
                 ]
     return cp.array(from_df_to_numpy(df))
 

python/cuml/tests/test_train_test_split.py

@@ -65,7 +65,7 @@ def test_split_dataframe(train_size, shuffle):
     y_reconstructed = cudf.concat([y_train, y_test]).sort_values()
 
     assert all(X_reconstructed.reset_index(drop=True) == X)
-    out = y_reconstructed.reset_index(drop=True).values_host == y.values_host
+    out = y_reconstructed.reset_index(drop=True).to_numpy() == y.to_numpy()
     assert all(out)