scikit-learn-intelex 2024.7.0__py311-none-win_amd64.whl → 2025.0.0__py311-none-win_amd64.whl

scikit_learn_intelex-2025.0.0.data/data/Lib/site-packages/sklearnex/linear_model/tests/test_incremental_ridge.py

@@ -0,0 +1,153 @@
+# ===============================================================================
+# Copyright 2024 Intel Corporation
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ===============================================================================
+
+from daal4py.sklearn._utils import daal_check_version
+
+if daal_check_version((2024, "P", 600)):
+    import numpy as np
+    import pytest
+    from numpy.testing import assert_allclose
+    from sklearn.exceptions import NotFittedError
+
+    from onedal.tests.utils._dataframes_support import (
+        _as_numpy,
+        _convert_to_dataframe,
+        get_dataframes_and_queues,
+    )
+    from sklearnex.linear_model import IncrementalRidge
+
+    def _compute_ridge_coefficients(X, y, alpha, fit_intercept):
+        coefficients_manual, intercept_manual = None, None
+        if fit_intercept:
+            X_mean = np.mean(X, axis=0)
+            y_mean = np.mean(y)
+            X_centered = X - X_mean
+            y_centered = y - y_mean
+
+            X_with_intercept = np.hstack([np.ones((X.shape[0], 1)), X_centered])
+            lambda_identity = alpha * np.eye(X_with_intercept.shape[1])
+            inverse_term = np.linalg.inv(
+                np.dot(X_with_intercept.T, X_with_intercept) + lambda_identity
+            )
+            xt_y = np.dot(X_with_intercept.T, y_centered)
+            coefficients_manual = np.dot(inverse_term, xt_y)
+
+            intercept_manual = y_mean - np.dot(X_mean, coefficients_manual[1:])
+            coefficients_manual = coefficients_manual[1:]
+        else:
+            lambda_identity = alpha * np.eye(X.shape[1])
+            inverse_term = np.linalg.inv(np.dot(X.T, X) + lambda_identity)
+            xt_y = np.dot(X.T, y)
+            coefficients_manual = np.dot(inverse_term, xt_y)
+
+        return coefficients_manual, intercept_manual
+
+    @pytest.mark.parametrize("dataframe,queue", get_dataframes_and_queues())
+    @pytest.mark.parametrize("batch_size", [10, 100, 1000])
+    @pytest.mark.parametrize("alpha", [0.1, 0.5, 1.0])
+    @pytest.mark.parametrize("fit_intercept", [True, False])
+    def test_inc_ridge_fit_coefficients(
+        dataframe, queue, alpha, batch_size, fit_intercept
+    ):
+        sample_size, feature_size = 1000, 50
+        X = np.random.rand(sample_size, feature_size)
+        y = np.random.rand(sample_size)
+        X_c = _convert_to_dataframe(X, sycl_queue=queue, target_df=dataframe)
+        y_c = _convert_to_dataframe(y, sycl_queue=queue, target_df=dataframe)
+
+        inc_ridge = IncrementalRidge(
+            fit_intercept=fit_intercept, alpha=alpha, batch_size=batch_size
+        )
+        inc_ridge.fit(X_c, y_c)
+
+        coefficients_manual, intercept_manual = _compute_ridge_coefficients(
+            X, y, alpha, fit_intercept
+        )
+        if fit_intercept:
+            assert_allclose(inc_ridge.intercept_, intercept_manual, rtol=1e-6, atol=1e-6)
+
+        assert_allclose(inc_ridge.coef_, coefficients_manual, rtol=1e-6, atol=1e-6)
+
+    @pytest.mark.parametrize("dataframe,queue", get_dataframes_and_queues())
+    @pytest.mark.parametrize("batch_size", [2, 5])
+    @pytest.mark.parametrize("alpha", [0.1, 0.5, 1.0])
+    def test_inc_ridge_partial_fit_coefficients(dataframe, queue, alpha, batch_size):
+        sample_size, feature_size = 1000, 50
+        X = np.random.rand(sample_size, feature_size)
+        y = np.random.rand(sample_size)
+        X_split = np.array_split(X, batch_size)
+        y_split = np.array_split(y, batch_size)
+
+        inc_ridge = IncrementalRidge(fit_intercept=False, alpha=alpha)
+
+        for batch_index in range(len(X_split)):
+            X_c = _convert_to_dataframe(
+                X_split[batch_index], sycl_queue=queue, target_df=dataframe
+            )
+            y_c = _convert_to_dataframe(
+                y_split[batch_index], sycl_queue=queue, target_df=dataframe
+            )
+            inc_ridge.partial_fit(X_c, y_c)
+
+        lambda_identity = alpha * np.eye(X.shape[1])
+        inverse_term = np.linalg.inv(np.dot(X.T, X) + lambda_identity)
+        xt_y = np.dot(X.T, y)
+        coefficients_manual = np.dot(inverse_term, xt_y)
+
+        assert_allclose(inc_ridge.coef_, coefficients_manual, rtol=1e-6, atol=1e-6)
+
+    def test_inc_ridge_score_before_fit():
+        X = np.array([[1, 1], [1, 2], [2, 2], [2, 3]])
+        y = np.dot(X, np.array([1, 2])) + 3
+        inc_ridge = IncrementalRidge(alpha=0.5)
+        with pytest.raises(NotFittedError):
+            inc_ridge.score(X, y)
+
+    def test_inc_ridge_predict_before_fit():
+        X = np.array([[1, 1], [1, 2], [2, 2], [2, 3]])
+        inc_ridge = IncrementalRidge(alpha=0.5)
+        with pytest.raises(NotFittedError):
+            inc_ridge.predict(X)
+
+    def test_inc_ridge_score_after_fit():
+        X = np.array([[1, 1], [1, 2], [2, 2], [2, 3]])
+        y = np.dot(X, np.array([1, 2])) + 3
+        inc_ridge = IncrementalRidge(alpha=0.5)
+        inc_ridge.fit(X, y)
+        assert inc_ridge.score(X, y) >= 0.97
+
+    @pytest.mark.parametrize("dataframe,queue", get_dataframes_and_queues())
+    @pytest.mark.parametrize("fit_intercept", [True, False])
+    def test_inc_ridge_predict_after_fit(dataframe, queue, fit_intercept):
+        sample_size, feature_size = 1000, 50
+        X = np.random.rand(sample_size, feature_size)
+        y = np.random.rand(sample_size)
+        X_c = _convert_to_dataframe(X, sycl_queue=queue, target_df=dataframe)
+        y_c = _convert_to_dataframe(y, sycl_queue=queue, target_df=dataframe)
+
+        inc_ridge = IncrementalRidge(fit_intercept=fit_intercept, alpha=0.5)
+        inc_ridge.fit(X_c, y_c)
+
+        y_pred = inc_ridge.predict(X_c)
+
+        coefficients_manual, intercept_manual = _compute_ridge_coefficients(
+            X, y, 0.5, fit_intercept
+        )
+        y_pred_manual = np.dot(X, coefficients_manual)
+        if fit_intercept:
+            y_pred_manual += intercept_manual
+
+        assert_allclose(_as_numpy(y_pred), y_pred_manual, rtol=1e-6, atol=1e-6)

{scikit_learn_intelex-2024.7.0.data → scikit_learn_intelex-2025.0.0.data}/data/Lib/site-packages/sklearnex/linear_model/tests/test_logreg.py

@@ -127,7 +127,8 @@ if daal_check_version((2024, "P", 700)):
             pred_sp = model_sp.predict(X_sp)
             prob_sp = model_sp.predict_proba(X_sp)
 
-        assert_allclose(pred, pred_sp)
-        assert_allclose(prob, prob_sp)
-        assert_allclose(model.coef_, model_sp.coef_, rtol=1e-4)
-        assert_allclose(model.intercept_, model_sp.intercept_, rtol=1e-4)
+        rtol = 2e-4
+        assert_allclose(pred, pred_sp, rtol=rtol)
+        assert_allclose(prob, prob_sp, rtol=rtol)
+        assert_allclose(model.coef_, model_sp.coef_, rtol=rtol)
+        assert_allclose(model.intercept_, model_sp.intercept_, rtol=rtol)

{scikit_learn_intelex-2024.7.0.data → scikit_learn_intelex-2025.0.0.data}/data/Lib/site-packages/sklearnex/manifold/t_sne.py

@@ -15,7 +15,7 @@
 # ===============================================================================
 
 from daal4py.sklearn.manifold import TSNE
-from onedal._device_offload import support_usm_ndarray
+from onedal._device_offload import support_input_format
 
-TSNE.fit = support_usm_ndarray(queue_param=False)(TSNE.fit)
-TSNE.fit_transform = support_usm_ndarray(queue_param=False)(TSNE.fit_transform)
+TSNE.fit = support_input_format(queue_param=False)(TSNE.fit)
+TSNE.fit_transform = support_input_format(queue_param=False)(TSNE.fit_transform)

{scikit_learn_intelex-2024.7.0.data → scikit_learn_intelex-2025.0.0.data}/data/Lib/site-packages/sklearnex/metrics/pairwise.py

@@ -15,8 +15,8 @@
 # ===============================================================================
 
 from daal4py.sklearn.metrics import pairwise_distances
-from onedal._device_offload import support_usm_ndarray
+from onedal._device_offload import support_input_format
 
-pairwise_distances = support_usm_ndarray(freefunc=True, queue_param=False)(
+pairwise_distances = support_input_format(freefunc=True, queue_param=False)(
     pairwise_distances
 )

{scikit_learn_intelex-2024.7.0.data → scikit_learn_intelex-2025.0.0.data}/data/Lib/site-packages/sklearnex/metrics/ranking.py

@@ -15,6 +15,6 @@
 # ===============================================================================
 
 from daal4py.sklearn.metrics import roc_auc_score
-from onedal._device_offload import support_usm_ndarray
+from onedal._device_offload import support_input_format
 
-roc_auc_score = support_usm_ndarray(freefunc=True, queue_param=False)(roc_auc_score)
+roc_auc_score = support_input_format(freefunc=True, queue_param=False)(roc_auc_score)

{scikit_learn_intelex-2024.7.0.data → scikit_learn_intelex-2025.0.0.data}/data/Lib/site-packages/sklearnex/model_selection/split.py

@@ -15,6 +15,8 @@
 # ===============================================================================
 
 from daal4py.sklearn.model_selection import train_test_split
-from onedal._device_offload import support_usm_ndarray
+from onedal._device_offload import support_input_format
 
-train_test_split = support_usm_ndarray(freefunc=True, queue_param=False)(train_test_split)
+train_test_split = support_input_format(freefunc=True, queue_param=False)(
+    train_test_split
+)

{scikit_learn_intelex-2024.7.0.data → scikit_learn_intelex-2025.0.0.data}/data/Lib/site-packages/sklearnex/neighbors/_lof.py

@@ -26,7 +26,8 @@ from daal4py.sklearn._utils import sklearn_check_version
 from sklearnex._device_offload import dispatch, wrap_output_data
 from sklearnex.neighbors.common import KNeighborsDispatchingBase
 from sklearnex.neighbors.knn_unsupervised import NearestNeighbors
-from sklearnex.utils import get_namespace
+
+from ..utils._array_api import get_namespace
 
 
 @control_n_jobs(decorated_methods=["fit", "_kneighbors"])

{scikit_learn_intelex-2024.7.0.data → scikit_learn_intelex-2025.0.0.data}/data/Lib/site-packages/sklearnex/neighbors/common.py

@@ -28,7 +28,7 @@ from daal4py.sklearn._utils import sklearn_check_version
 from onedal.utils import _check_array, _num_features, _num_samples
 
 from .._utils import PatchingConditionsChain
-from ..utils import get_namespace
+from ..utils._array_api import get_namespace
 
 
 class KNeighborsDispatchingBase:

{scikit_learn_intelex-2024.7.0.data/data/Lib/site-packages/sklearnex/preview/cluster → scikit_learn_intelex-2025.0.0.data/data/Lib/site-packages/sklearnex/preview}/__init__.py

@@ -14,6 +14,4 @@
 # limitations under the License.
 # ==============================================================================
 
-from .k_means import KMeans
-
-__all__ = ["KMeans"]
+__all__ = ["covariance", "decomposition", "linear_model"]

{scikit_learn_intelex-2024.7.0.data → scikit_learn_intelex-2025.0.0.data}/data/Lib/site-packages/sklearnex/preview/decomposition/incremental_pca.py

@@ -61,7 +61,7 @@ class IncrementalPCA(sklearn_IncrementalPCA):
         return self._onedal_transform(X, queue)
 
     def _onedal_partial_fit(self, X, check_input=True, queue=None):
-        first_pass = not hasattr(self, "components_")
+        first_pass = not hasattr(self, "_onedal_estimator")
 
         if check_input:
             if sklearn_check_version("1.0"):
@@ -78,10 +78,10 @@ class IncrementalPCA(sklearn_IncrementalPCA):
         n_samples, n_features = X.shape
 
         if self.n_components is None:
-            if not hasattr(self, "components_"):
+            if not hasattr(self, "_components_shape"):
                 self.n_components_ = min(n_samples, n_features)
-            else:
-                self.n_components_ = self.components_.shape[0]
+                self._components_shape = self.n_components_
+
         elif not self.n_components <= n_features:
             raise ValueError(
                 "n_components=%r invalid for n_features=%d, need "
@@ -106,12 +106,12 @@ class IncrementalPCA(sklearn_IncrementalPCA):
 
         if not hasattr(self, "_onedal_estimator"):
             self._onedal_estimator = self._onedal_incremental_pca(**onedal_params)
-        self._onedal_estimator.partial_fit(X, queue)
+        self._onedal_estimator.partial_fit(X, queue=queue)
         self._need_to_finalize = True
 
-    def _onedal_finalize_fit(self):
+    def _onedal_finalize_fit(self, queue=None):
         assert hasattr(self, "_onedal_estimator")
-        self._onedal_estimator.finalize_fit()
+        self._onedal_estimator.finalize_fit(queue=queue)
         self._need_to_finalize = False
 
     def _onedal_fit(self, X, queue=None):
@@ -142,7 +142,7 @@ class IncrementalPCA(sklearn_IncrementalPCA):
             X_batch = X[batch]
             self._onedal_partial_fit(X_batch, queue=queue)
 
-        self._onedal_finalize_fit()
+        self._onedal_finalize_fit(queue=queue)
 
         return self
 

{scikit_learn_intelex-2024.7.0.data → scikit_learn_intelex-2025.0.0.data}/data/Lib/site-packages/sklearnex/preview/decomposition/tests/test_incremental_pca.py

@@ -74,7 +74,7 @@ def check_pca_on_gold_data(incpca, dtype, whiten, transformed_data):
     )
 
     tol = 1e-7
-    if transformed_data.dtype == np.float32:
+    if dtype == np.float32:
         tol = 7e-6 if whiten else 1e-6
 
     assert incpca.n_samples_seen_ == expected_n_samples_seen_
@@ -112,7 +112,7 @@ def check_pca_on_gold_data(incpca, dtype, whiten, transformed_data):
 
 
 def check_pca(incpca, dtype, whiten, data, transformed_data):
-    tol = 3e-3 if transformed_data.dtype == np.float32 else 2e-6
+    tol = 3e-3 if dtype == np.float32 else 2e-6
 
     n_components = incpca.n_components_
 

{scikit_learn_intelex-2024.7.0.data → scikit_learn_intelex-2025.0.0.data}/data/Lib/site-packages/sklearnex/preview/linear_model/ridge.py

@@ -408,11 +408,11 @@ if daal_check_version((2024, "P", 600)):
 
 else:
     from daal4py.sklearn.linear_model._ridge import Ridge
-    from onedal._device_offload import support_usm_ndarray
+    from onedal._device_offload import support_input_format
 
-    Ridge.fit = support_usm_ndarray(queue_param=False)(Ridge.fit)
-    Ridge.predict = support_usm_ndarray(queue_param=False)(Ridge.predict)
-    Ridge.score = support_usm_ndarray(queue_param=False)(Ridge.score)
+    Ridge.fit = support_input_format(queue_param=False)(Ridge.fit)
+    Ridge.predict = support_input_format(queue_param=False)(Ridge.predict)
+    Ridge.score = support_input_format(queue_param=False)(Ridge.score)
 
     logging.warning(
         "Preview Ridge requires oneDAL version >= 2024.6 but it was not found"

scikit_learn_intelex-2025.0.0.data/data/Lib/site-packages/sklearnex/spmd/basic_statistics/__init__.py

@@ -0,0 +1,20 @@
+# ==============================================================================
+# Copyright 2023 Intel Corporation
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ==============================================================================
+
+from .basic_statistics import BasicStatistics
+from .incremental_basic_statistics import IncrementalBasicStatistics
+
+__all__ = ["BasicStatistics", "IncrementalBasicStatistics"]

scikit_learn_intelex-2025.0.0.data/data/Lib/site-packages/sklearnex/spmd/basic_statistics/incremental_basic_statistics.py

@@ -0,0 +1,30 @@
+# ==============================================================================
+# Copyright 2024 Intel Corporation
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ==============================================================================
+
+
+from onedal.spmd.basic_statistics import (
+    IncrementalBasicStatistics as onedalSPMD_IncrementalBasicStatistics,
+)
+
+from ...basic_statistics import (
+    IncrementalBasicStatistics as base_IncrementalBasicStatistics,
+)
+
+
+class IncrementalBasicStatistics(base_IncrementalBasicStatistics):
+    _onedal_incremental_basic_statistics = staticmethod(
+        onedalSPMD_IncrementalBasicStatistics
+    )