scikit-learn-intelex 2024.5.0__py311-none-win_amd64.whl → 2024.7.0__py311-none-win_amd64.whl

scikit_learn_intelex-2024.7.0.data/data/Lib/site-packages/sklearnex/preview/linear_model/ridge.py

@@ -0,0 +1,419 @@
+# ===============================================================================
+# 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.
+# ===============================================================================
+
+import logging
+
+from daal4py.sklearn._utils import daal_check_version, sklearn_check_version
+
+if daal_check_version((2024, "P", 600)):
+    import numbers
+
+    import numpy as np
+    from scipy.sparse import issparse
+    from sklearn.linear_model import Ridge as sklearn_Ridge
+    from sklearn.metrics import r2_score
+    from sklearn.utils.validation import check_is_fitted, check_X_y
+
+    from daal4py.sklearn.linear_model._ridge import _fit_ridge as daal4py_fit_ridge
+
+    if sklearn_check_version("1.0") and not sklearn_check_version("1.2"):
+        from sklearn.linear_model._base import _deprecate_normalize
+    if sklearn_check_version("1.1") and not sklearn_check_version("1.2"):
+        from sklearn.utils import check_scalar
+
+    from onedal.linear_model import Ridge as onedal_Ridge
+    from onedal.utils import _num_features, _num_samples
+
+    from ..._device_offload import dispatch, wrap_output_data
+    from ..._utils import PatchingConditionsChain
+
+    def _is_numeric_scalar(value):
+        """
+        Determines if the provided value is a numeric scalar.
+
+        Args:
+        value: The value to be checked.
+
+        Returns:
+        bool: True if the value is a numeric scalar, False otherwise.
+        """
+        return isinstance(value, numbers.Real)
+
+    class Ridge(sklearn_Ridge):
+        __doc__ = sklearn_Ridge.__doc__
+
+        if sklearn_check_version("1.2"):
+            _parameter_constraints: dict = {**sklearn_Ridge._parameter_constraints}
+
+            def __init__(
+                self,
+                alpha=1.0,
+                fit_intercept=True,
+                copy_X=True,
+                max_iter=None,
+                tol=1e-4,
+                solver="auto",
+                positive=False,
+                random_state=None,
+            ):
+                super().__init__(
+                    alpha=alpha,
+                    fit_intercept=fit_intercept,
+                    copy_X=copy_X,
+                    max_iter=max_iter,
+                    tol=tol,
+                    solver=solver,
+                    positive=positive,
+                    random_state=random_state,
+                )
+
+        elif sklearn_check_version("1.0"):
+
+            def __init__(
+                self,
+                alpha=1.0,
+                fit_intercept=True,
+                normalize="deprecated",
+                copy_X=True,
+                max_iter=None,
+                tol=1e-3,
+                solver="auto",
+                positive=False,
+                random_state=None,
+            ):
+                super().__init__(
+                    alpha=alpha,
+                    fit_intercept=fit_intercept,
+                    normalize=normalize,
+                    copy_X=copy_X,
+                    max_iter=max_iter,
+                    solver=solver,
+                    tol=tol,
+                    positive=positive,
+                    random_state=random_state,
+                )
+
+        else:
+
+            def __init__(
+                self,
+                alpha=1.0,
+                fit_intercept=True,
+                normalize=False,
+                copy_X=True,
+                max_iter=None,
+                tol=1e-3,
+                solver="auto",
+                random_state=None,
+            ):
+                super().__init__(
+                    alpha=alpha,
+                    fit_intercept=fit_intercept,
+                    normalize=normalize,
+                    copy_X=copy_X,
+                    max_iter=max_iter,
+                    tol=tol,
+                    solver=solver,
+                    random_state=random_state,
+                )
+
+        def fit(self, X, y, sample_weight=None):
+            # It is necessary to properly update coefs for predict if we
+            # fallback to sklearn in dispatch
+            if hasattr(self, "_onedal_estimator"):
+                del self._onedal_estimator
+
+            dispatch(
+                self,
+                "fit",
+                {
+                    "onedal": self.__class__._onedal_fit,
+                    "sklearn": sklearn_Ridge.fit,
+                },
+                X,
+                y,
+                sample_weight,
+            )
+            return self
+
+        @wrap_output_data
+        def predict(self, X):
+            check_is_fitted(self)
+
+            return dispatch(
+                self,
+                "predict",
+                {
+                    "onedal": self.__class__._onedal_predict,
+                    "sklearn": sklearn_Ridge.predict,
+                },
+                X,
+            )
+
+        @wrap_output_data
+        def score(self, X, y, sample_weight=None):
+            check_is_fitted(self)
+
+            return dispatch(
+                self,
+                "score",
+                {
+                    "onedal": self.__class__._onedal_score,
+                    "sklearn": sklearn_Ridge.score,
+                },
+                X,
+                y,
+                sample_weight=sample_weight,
+            )
+
+        def _onedal_fit_supported(self, patching_status, method_name, *data):
+            assert method_name == "fit"
+            assert len(data) == 3
+            X, y, sample_weight = data
+
+            normalize_is_set = (
+                hasattr(self, "normalize")
+                and self.normalize
+                and self.normalize != "deprecated"
+            )
+            positive_is_set = hasattr(self, "positive") and self.positive
+
+            n_samples = _num_samples(X)
+            n_features = _num_features(X, fallback_1d=True)
+
+            # Check if equations are well defined
+            is_underdetermined = n_samples < (n_features + int(self.fit_intercept))
+
+            patching_status.and_conditions(
+                [
+                    (
+                        self.solver == "auto",
+                        f"'{self.solver}' solver is not supported. "
+                        "Only 'auto' solver is supported.",
+                    ),
+                    (
+                        not issparse(X) and not issparse(y),
+                        "Sparse input is not supported.",
+                    ),
+                    (
+                        not is_underdetermined,
+                        "The shape of X (fitting) does not satisfy oneDAL requirements:"
+                        "Number of features + 1 >= number of samples.",
+                    ),
+                    (sample_weight is None, "Sample weight is not supported."),
+                    (not normalize_is_set, "Normalization is not supported."),
+                    (
+                        not positive_is_set,
+                        "Forced positive coefficients are not supported.",
+                    ),
+                ]
+            )
+
+            return patching_status
+
+        def _onedal_predict_supported(self, patching_status, method_name, *data):
+            assert method_name in ["predict", "score"]
+            assert len(data) <= 2
+
+            n_samples = _num_samples(data[0])
+            model_is_sparse = issparse(self.coef_) or (
+                self.fit_intercept and issparse(self.intercept_)
+            )
+            patching_status.and_conditions(
+                [
+                    (
+                        self.solver == "auto",
+                        f"'{self.solver}' solver is not supported. "
+                        "Only 'auto' solver is supported.",
+                    ),
+                    (n_samples > 0, "Number of samples is less than 1."),
+                    (not issparse(data[0]), "Sparse input is not supported."),
+                    (not model_is_sparse, "Sparse coefficients are not supported."),
+                ]
+            )
+
+            return patching_status
+
+        def _onedal_gpu_supported(self, method_name, *data):
+            patching_status = PatchingConditionsChain(
+                f"sklearn.linear_model.{self.__class__.__name__}.fit"
+            )
+
+            if method_name == "fit":
+                patching_status.and_condition(
+                    _is_numeric_scalar(self.alpha),
+                    "Non-scalar alpha is not supported for GPU.",
+                )
+
+                return self._onedal_fit_supported(patching_status, method_name, *data)
+
+            if method_name in ["predict", "score"]:
+                return self._onedal_predict_supported(patching_status, method_name, *data)
+
+            raise RuntimeError(
+                f"Unknown method {method_name} in {self.__class__.__name__}"
+            )
+
+        def _onedal_cpu_supported(self, method_name, *data):
+            patching_status = PatchingConditionsChain(
+                f"sklearn.linear_model.{self.__class__.__name__}.fit"
+            )
+
+            if method_name == "fit":
+                return self._onedal_fit_supported(patching_status, method_name, *data)
+
+            if method_name in ["predict", "score"]:
+                return self._onedal_predict_supported(patching_status, method_name, *data)
+
+            raise RuntimeError(
+                f"Unknown method {method_name} in {self.__class__.__name__}"
+            )
+
+        def _initialize_onedal_estimator(self):
+            onedal_params = {
+                "fit_intercept": self.fit_intercept,
+                "alpha": self.alpha,
+                "copy_X": self.copy_X,
+            }
+            self._onedal_estimator = onedal_Ridge(**onedal_params)
+
+        def _daal_fit(self, X, y, sample_weight=None):
+            daal4py_fit_ridge(self, X, y, sample_weight)
+            self._onedal_estimator.n_features_in_ = _num_features(X, fallback_1d=True)
+            self._onedal_estimator.coef_ = self.coef_
+            self._onedal_estimator.intercept_ = self.intercept_
+
+        def _onedal_fit(self, X, y, sample_weight, queue=None):
+            # `Sample weight` is not supported. Expected to be None value.
+            assert sample_weight is None
+
+            if sklearn_check_version("1.2"):
+                self._validate_params()
+            elif sklearn_check_version("1.1"):
+                if self.max_iter is not None:
+                    self.max_iter = check_scalar(
+                        self.max_iter, "max_iter", target_type=numbers.Integral, min_val=1
+                    )
+                self.tol = check_scalar(
+                    self.tol, "tol", target_type=numbers.Real, min_val=0.0
+                )
+                if self.alpha is not None and not isinstance(
+                    self.alpha, (np.ndarray, tuple)
+                ):
+                    self.alpha = check_scalar(
+                        self.alpha,
+                        "alpha",
+                        target_type=numbers.Real,
+                        min_val=0.0,
+                        include_boundaries="left",
+                    )
+
+            check_params = {
+                "X": X,
+                "y": y,
+                "dtype": [np.float64, np.float32],
+                "accept_sparse": ["csr", "csc", "coo"],
+                "y_numeric": True,
+                "multi_output": True,
+            }
+            if sklearn_check_version("1.0"):
+                X, y = self._validate_data(**check_params)
+            else:
+                X, y = check_X_y(**check_params)
+
+            if sklearn_check_version("1.0") and not sklearn_check_version("1.2"):
+                self._normalize = _deprecate_normalize(
+                    self.normalize,
+                    default=False,
+                    estimator_name=self.__class__.__name__,
+                )
+
+            self._initialize_onedal_estimator()
+
+            # Falling back to daal4py if the device is CPU and alpha is array-like
+            # since onedal does not yet support non-scalars for alpha, thus
+            # should only be used for GPU/CPU with scalar alpha to not limit the functionality
+            cpu_device = queue is None or queue.sycl_device.is_cpu
+            if cpu_device and not _is_numeric_scalar(self.alpha):
+                self._daal_fit(X, y)
+            else:
+                self._onedal_estimator.fit(X, y, queue=queue)
+
+            self._save_attributes()
+
+        def _onedal_predict(self, X, queue=None):
+            if sklearn_check_version("1.0"):
+                X = self._validate_data(X, accept_sparse=False, reset=False)
+
+            if not hasattr(self, "_onedal_estimator"):
+                self._initialize_onedal_estimator()
+                self._onedal_estimator.coef_ = self.coef_
+                self._onedal_estimator.intercept_ = self.intercept_
+
+            res = self._onedal_estimator.predict(X, queue=queue)
+            return res
+
+        def _onedal_score(self, X, y, sample_weight=None, queue=None):
+            return r2_score(
+                y, self._onedal_predict(X, queue=queue), sample_weight=sample_weight
+            )
+
+        @property
+        def coef_(self):
+            return self._coef
+
+        @coef_.setter
+        def coef_(self, value):
+            if hasattr(self, "_onedal_estimator"):
+                self._onedal_estimator.coef_ = value
+                # checking if the model is already fitted and if so, deleting the model
+                if hasattr(self._onedal_estimator, "_onedal_model"):
+                    del self._onedal_estimator._onedal_model
+            self._coef = value
+
+        @property
+        def intercept_(self):
+            return self._intercept
+
+        @intercept_.setter
+        def intercept_(self, value):
+            if hasattr(self, "_onedal_estimator"):
+                self._onedal_estimator.intercept_ = value
+                # checking if the model is already fitted and if so, deleting the model
+                if hasattr(self._onedal_estimator, "_onedal_model"):
+                    del self._onedal_estimator._onedal_model
+            self._intercept = value
+
+        def _save_attributes(self):
+            self.n_features_in_ = self._onedal_estimator.n_features_in_
+            self._coef = self._onedal_estimator.coef_
+            self._intercept = self._onedal_estimator.intercept_
+
+        fit.__doc__ = sklearn_Ridge.fit.__doc__
+        predict.__doc__ = sklearn_Ridge.predict.__doc__
+        score.__doc__ = sklearn_Ridge.score.__doc__
+
+else:
+    from daal4py.sklearn.linear_model._ridge import Ridge
+    from onedal._device_offload import support_usm_ndarray
+
+    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)
+
+    logging.warning(
+        "Preview Ridge requires oneDAL version >= 2024.6 but it was not found"
+    )

scikit_learn_intelex-2024.7.0.data/data/Lib/site-packages/sklearnex/preview/linear_model/tests/test_ridge.py

@@ -0,0 +1,102 @@
+# ===============================================================================
+# 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.
+# ===============================================================================
+
+import numpy
+import pytest
+from numpy.testing import assert_allclose
+from sklearn.exceptions import NotFittedError
+
+from daal4py.sklearn._utils import daal_check_version
+from onedal.tests.utils._dataframes_support import (
+    _convert_to_dataframe,
+    get_dataframes_and_queues,
+)
+
+
+@pytest.mark.parametrize("dataframe,queue", get_dataframes_and_queues())
+def test_sklearnex_import_ridge(dataframe, queue):
+    from sklearnex.preview.linear_model import Ridge
+
+    X = numpy.array([[1, 1], [1, 2], [2, 2], [2, 3]])
+    y = numpy.dot(X, numpy.array([1, 2])) + 3
+    X_c = _convert_to_dataframe(X, sycl_queue=queue, target_df=dataframe)
+    y_c = _convert_to_dataframe(y, sycl_queue=queue, target_df=dataframe)
+    ridge_reg = Ridge(alpha=0.5).fit(X_c, y_c)
+
+    if daal_check_version((2024, "P", 600)):
+        assert "preview" in ridge_reg.__module__
+    else:
+        assert "daal4py" in ridge_reg.__module__
+
+    assert_allclose(ridge_reg.intercept_, 3.86, rtol=1e-2)
+    assert_allclose(ridge_reg.coef_, [0.91, 1.64], rtol=1e-2)
+
+
+@pytest.mark.parametrize("dataframe,queue", get_dataframes_and_queues())
+@pytest.mark.parametrize("sample_size", [100, 1000])
+@pytest.mark.parametrize("feature_size", [10, 50])
+@pytest.mark.parametrize("alpha", [0.1, 0.5, 1.0])
+def test_ridge_coefficients(dataframe, queue, sample_size, feature_size, alpha):
+    from sklearnex.preview.linear_model import Ridge
+
+    X = numpy.random.rand(sample_size, feature_size)
+    y = numpy.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)
+    ridge_reg = Ridge(fit_intercept=False, alpha=alpha).fit(X_c, y_c)
+
+    # computing the coefficients manually
+    # using the normal equation formula: (X^T * X + lambda * I)^-1 * X^T * y
+    lambda_identity = alpha * numpy.eye(X.shape[1])
+    inverse_term = numpy.linalg.inv(numpy.dot(X.T, X) + lambda_identity)
+    xt_y = numpy.dot(X.T, y)
+    coefficients_manual = numpy.dot(inverse_term, xt_y)
+
+    assert_allclose(ridge_reg.coef_, coefficients_manual, rtol=1e-6, atol=1e-6)
+
+
+if daal_check_version((2024, "P", 600)):
+
+    @pytest.mark.parametrize("dataframe,queue", get_dataframes_and_queues())
+    def test_ridge_score_before_fit(dataframe, queue):
+        from sklearnex.preview.linear_model import Ridge
+
+        sample_count, feature_count = 10, 5
+
+        model = Ridge(fit_intercept=True, alpha=0.5)
+
+        X, y = numpy.random.rand(sample_count, feature_count), numpy.random.rand(
+            sample_count
+        )
+        X_c = _convert_to_dataframe(X, sycl_queue=queue, target_df=dataframe)
+        y_c = _convert_to_dataframe(y, sycl_queue=queue, target_df=dataframe)
+
+        with pytest.raises(NotFittedError):
+            model.score(X_c, y_c)
+
+    @pytest.mark.parametrize("dataframe,queue", get_dataframes_and_queues())
+    def test_ridge_predict_before_fit(dataframe, queue):
+        from sklearnex.preview.linear_model import Ridge
+
+        sample_count, feature_count = 10, 5
+
+        model = Ridge(fit_intercept=True, alpha=0.5)
+
+        X = numpy.random.rand(sample_count, feature_count)
+        X_c = _convert_to_dataframe(X, sycl_queue=queue, target_df=dataframe)
+
+        with pytest.raises(NotFittedError):
+            model.predict(X_c)

scikit_learn_intelex-2024.7.0.data/data/Lib/site-packages/sklearnex/spmd/basic_statistics/tests/test_basic_statistics_spmd.py

@@ -0,0 +1,107 @@
+# ==============================================================================
+# 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.
+# ==============================================================================
+
+import numpy as np
+import pytest
+from numpy.testing import assert_allclose
+
+from onedal.basic_statistics.tests.test_basic_statistics import options_and_tests
+from onedal.tests.utils._dataframes_support import (
+    _convert_to_dataframe,
+    get_dataframes_and_queues,
+)
+from sklearnex.tests._utils_spmd import (
+    _generate_statistic_data,
+    _get_local_tensor,
+    _mpi_libs_and_gpu_available,
+)
+
+
+@pytest.mark.skipif(
+    not _mpi_libs_and_gpu_available,
+    reason="GPU device and MPI libs required for test",
+)
+@pytest.mark.parametrize(
+    "dataframe,queue",
+    get_dataframes_and_queues(dataframe_filter_="dpnp,dpctl", device_filter_="gpu"),
+)
+@pytest.mark.mpi
+def test_basic_stats_spmd_gold(dataframe, queue):
+    # Import spmd and batch algo
+    from onedal.basic_statistics import BasicStatistics as BasicStatistics_Batch
+    from sklearnex.spmd.basic_statistics import BasicStatistics as BasicStatistics_SPMD
+
+    # Create gold data and convert to dataframe
+    data = np.array(
+        [
+            [0.0, 0.0, 0.0],
+            [0.0, 1.0, 2.0],
+            [0.0, 2.0, 4.0],
+            [0.0, 3.0, 8.0],
+            [0.0, 4.0, 16.0],
+            [0.0, 5.0, 32.0],
+            [0.0, 6.0, 64.0],
+        ]
+    )
+
+    local_dpt_data = _convert_to_dataframe(
+        _get_local_tensor(data), sycl_queue=queue, target_df=dataframe
+    )
+
+    # Ensure results of batch algo match spmd
+    spmd_result = BasicStatistics_SPMD().fit(local_dpt_data)
+    batch_result = BasicStatistics_Batch().fit(data)
+
+    for option in (opt[0] for opt in options_and_tests):
+        assert_allclose(getattr(spmd_result, option), getattr(batch_result, option))
+
+
+@pytest.mark.skipif(
+    not _mpi_libs_and_gpu_available,
+    reason="GPU device and MPI libs required for test",
+)
+@pytest.mark.parametrize("n_samples", [100, 10000])
+@pytest.mark.parametrize("n_features", [10, 100])
+@pytest.mark.parametrize("dtype", [np.float32, np.float64])
+@pytest.mark.parametrize(
+    "dataframe,queue",
+    get_dataframes_and_queues(dataframe_filter_="dpnp,dpctl", device_filter_="gpu"),
+)
+@pytest.mark.mpi
+def test_basic_stats_spmd_synthetic(n_samples, n_features, dataframe, queue, dtype):
+    # Import spmd and batch algo
+    from onedal.basic_statistics import BasicStatistics as BasicStatistics_Batch
+    from sklearnex.spmd.basic_statistics import BasicStatistics as BasicStatistics_SPMD
+
+    # Generate data and convert to dataframe
+    data = _generate_statistic_data(n_samples, n_features, dtype=dtype)
+
+    local_dpt_data = _convert_to_dataframe(
+        _get_local_tensor(data), sycl_queue=queue, target_df=dataframe
+    )
+
+    # Ensure results of batch algo match spmd
+    spmd_result = BasicStatistics_SPMD().fit(local_dpt_data)
+    batch_result = BasicStatistics_Batch().fit(data)
+
+    tol = 1e-5 if dtype == np.float32 else 1e-7
+    for option in (opt[0] for opt in options_and_tests):
+        assert_allclose(
+            getattr(spmd_result, option),
+            getattr(batch_result, option),
+            atol=tol,
+            rtol=tol,
+        )