scikit-learn-intelex 2024.4.0__py312-none-win_amd64.whl → 2025.10.0__py312-none-win_amd64.whl

scikit_learn_intelex-2025.10.0.data/data/Lib/site-packages/sklearnex/preview/decomposition/tests/test_incremental_pca.py

@@ -0,0 +1,390 @@
+# ===============================================================================
+# 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
+import scipy.sparse as sp
+from numpy.testing import assert_allclose
+from sklearn.base import clone
+from sklearn.datasets import load_iris
+
+from daal4py.sklearn._utils import daal_check_version, sklearn_check_version
+from onedal.tests.utils._dataframes_support import (
+    _as_numpy,
+    _convert_to_dataframe,
+    get_dataframes_and_queues,
+)
+from sklearnex.preview.decomposition import IncrementalPCA
+
+
+@pytest.mark.parametrize("dataframe,queue", get_dataframes_and_queues())
+def test_sklearnex_import(dataframe, queue):
+    X = [[-1, -1], [-2, -1], [-3, -2], [1, 1], [2, 1], [3, 2]]
+    X = _convert_to_dataframe(X, sycl_queue=queue, target_df=dataframe)
+    incpca = IncrementalPCA(n_components=2)
+    result = incpca.fit(X)
+    assert "sklearnex" in incpca.__module__
+    assert hasattr(incpca, "_onedal_estimator")
+    assert_allclose(_as_numpy(result.singular_values_), [6.30061232, 0.54980396])
+
+
+def check_pca_on_gold_data(incpca, dtype, whiten, transformed_data):
+    expected_n_samples_seen_ = 6
+    expected_n_features_in_ = 2
+    expected_n_components_ = 2
+    expected_components_ = np.array([[0.83849224, 0.54491354], [-0.54491354, 0.83849224]])
+    expected_singular_values_ = np.array([6.30061232, 0.54980396])
+    expected_mean_ = np.array([0, 0])
+    expected_var_ = np.array([5.6, 2.4])
+    expected_explained_variance_ = np.array([7.93954312, 0.06045688])
+    expected_explained_variance_ratio_ = np.array([0.99244289, 0.00755711])
+    expected_noise_variance_ = 0.0
+    expected_transformed_data = (
+        np.array(
+            [
+                [-0.49096647, -1.19399271],
+                [-0.78854479, 1.02218579],
+                [-1.27951125, -0.17180692],
+                [0.49096647, 1.19399271],
+                [0.78854479, -1.02218579],
+                [1.27951125, 0.17180692],
+            ]
+        )
+        if whiten
+        else np.array(
+            [
+                [-1.38340578, -0.2935787],
+                [-2.22189802, 0.25133484],
+                [-3.6053038, -0.04224385],
+                [1.38340578, 0.2935787],
+                [2.22189802, -0.25133484],
+                [3.6053038, 0.04224385],
+            ]
+        )
+    )
+
+    tol = 1e-7
+    if dtype == np.float32:
+        tol = 7e-6 if whiten else 1e-6
+
+    assert incpca.n_samples_seen_ == expected_n_samples_seen_
+    assert incpca.n_features_in_ == expected_n_features_in_
+    assert incpca.n_components_ == expected_n_components_
+
+    assert_allclose(incpca.singular_values_, expected_singular_values_, atol=tol)
+    assert_allclose(incpca.mean_, expected_mean_, atol=tol)
+    assert_allclose(incpca.var_, expected_var_, atol=tol)
+    assert_allclose(incpca.explained_variance_, expected_explained_variance_, atol=tol)
+    assert_allclose(
+        incpca.explained_variance_ratio_, expected_explained_variance_ratio_, atol=tol
+    )
+    assert np.abs(incpca.noise_variance_ - expected_noise_variance_) < tol
+    if daal_check_version((2024, "P", 500)):
+        assert_allclose(incpca.components_, expected_components_, atol=tol)
+        assert_allclose(_as_numpy(transformed_data), expected_transformed_data, atol=tol)
+    else:
+        for i in range(incpca.n_components_):
+            abs_dot_product = np.abs(
+                np.dot(incpca.components_[i], expected_components_[i])
+            )
+            assert np.abs(abs_dot_product - 1.0) < tol
+
+            if np.dot(incpca.components_[i], expected_components_[i]) < 0:
+                assert_allclose(
+                    _as_numpy(-transformed_data[i]),
+                    expected_transformed_data[i],
+                    atol=tol,
+                )
+            else:
+                assert_allclose(
+                    _as_numpy(transformed_data[i]), expected_transformed_data[i], atol=tol
+                )
+
+
+def check_pca(incpca, dtype, whiten, data, transformed_data):
+    tol = 3e-3 if dtype == np.float32 else 2e-6
+
+    n_components = incpca.n_components_
+
+    expected_n_samples_seen = data.shape[0]
+    expected_n_features_in = data.shape[1]
+    n_samples_seen = incpca.n_samples_seen_
+    n_features_in = incpca.n_features_in_
+    assert n_samples_seen == expected_n_samples_seen
+    assert n_features_in == expected_n_features_in
+
+    components = incpca.components_
+    singular_values = incpca.singular_values_
+    centered_data = data - np.mean(data, axis=0)
+    cov_eigenvalues, cov_eigenvectors = np.linalg.eig(
+        centered_data.T @ centered_data / (n_samples_seen - 1)
+    )
+    cov_eigenvalues = np.nan_to_num(cov_eigenvalues)
+    cov_eigenvalues[cov_eigenvalues < 0] = 0
+    eigenvalues_order = np.argsort(cov_eigenvalues)[::-1]
+    sorted_eigenvalues = cov_eigenvalues[eigenvalues_order]
+    sorted_eigenvectors = cov_eigenvectors[:, eigenvalues_order]
+    expected_singular_values = np.sqrt(sorted_eigenvalues * (n_samples_seen - 1))[
+        :n_components
+    ]
+    expected_components = sorted_eigenvectors.T[:n_components]
+
+    assert_allclose(singular_values, expected_singular_values, atol=tol)
+    for i in range(n_components):
+        component_length = np.dot(components[i], components[i])
+        assert np.abs(component_length - 1.0) < tol
+        abs_dot_product = np.abs(np.dot(components[i], expected_components[i]))
+        assert np.abs(abs_dot_product - 1.0) < tol
+
+    expected_mean = np.mean(data, axis=0)
+    assert_allclose(incpca.mean_, expected_mean, atol=tol)
+
+    expected_var = np.var(_as_numpy(data), ddof=1, axis=0)
+    assert_allclose(incpca.var_, expected_var, atol=tol)
+
+    expected_explained_variance = sorted_eigenvalues[:n_components]
+    assert_allclose(incpca.explained_variance_, expected_explained_variance, atol=tol)
+
+    expected_explained_variance_ratio = expected_explained_variance / np.sum(
+        sorted_eigenvalues
+    )
+    assert_allclose(
+        incpca.explained_variance_ratio_, expected_explained_variance_ratio, atol=tol
+    )
+
+    expected_noise_variance = (
+        np.mean(sorted_eigenvalues[n_components:])
+        if len(sorted_eigenvalues) > n_components
+        else 0.0
+    )
+    # TODO Fix noise variance computation (It is necessary to update C++ side)
+    # assert np.abs(incpca.noise_variance_ - expected_noise_variance) < tol
+
+    expected_transformed_data = centered_data @ components.T
+    if whiten:
+        scale = np.sqrt(incpca.explained_variance_)
+        min_scale = np.finfo(scale.dtype).eps
+        scale[scale < min_scale] = np.inf
+        expected_transformed_data /= scale
+
+    if not (whiten and n_components == n_samples_seen):
+        assert_allclose(_as_numpy(transformed_data), expected_transformed_data, atol=tol)
+
+
+@pytest.mark.parametrize("dataframe,queue", get_dataframes_and_queues())
+@pytest.mark.parametrize("whiten", [True, False])
+@pytest.mark.parametrize("num_blocks", [1, 2, 3])
+@pytest.mark.parametrize("dtype", [np.float32, np.float64])
+def test_sklearnex_partial_fit_on_gold_data(dataframe, queue, whiten, num_blocks, dtype):
+
+    X = np.array([[-1, -1], [-2, -1], [-3, -2], [1, 1], [2, 1], [3, 2]])
+    X = X.astype(dtype=dtype)
+    X_split = np.array_split(X, num_blocks)
+    incpca = IncrementalPCA(whiten=whiten)
+
+    for i in range(num_blocks):
+        X_split_df = _convert_to_dataframe(
+            X_split[i], sycl_queue=queue, target_df=dataframe
+        )
+        incpca.partial_fit(X_split_df)
+
+    X_df = _convert_to_dataframe(X, sycl_queue=queue, target_df=dataframe)
+    transformed_data = incpca.transform(X_df)
+    check_pca_on_gold_data(incpca, dtype, whiten, transformed_data)
+
+
+@pytest.mark.parametrize("dataframe,queue", get_dataframes_and_queues())
+@pytest.mark.parametrize("whiten", [True, False])
+@pytest.mark.parametrize("num_blocks", [1, 2, 3])
+@pytest.mark.parametrize("dtype", [np.float32, np.float64])
+def test_sklearnex_fit_on_gold_data(dataframe, queue, whiten, num_blocks, dtype):
+
+    X = np.array([[-1, -1], [-2, -1], [-3, -2], [1, 1], [2, 1], [3, 2]])
+    X = X.astype(dtype=dtype)
+    incpca = IncrementalPCA(whiten=whiten, batch_size=X.shape[0] // num_blocks)
+
+    X_df = _convert_to_dataframe(X, sycl_queue=queue, target_df=dataframe)
+    incpca.fit(X_df)
+    transformed_data = incpca.transform(X_df)
+
+    check_pca_on_gold_data(incpca, dtype, whiten, transformed_data)
+
+
+@pytest.mark.parametrize("dataframe,queue", get_dataframes_and_queues())
+@pytest.mark.parametrize("whiten", [True, False])
+@pytest.mark.parametrize("num_blocks", [1, 2, 3])
+@pytest.mark.parametrize("dtype", [np.float32, np.float64])
+def test_sklearnex_fit_transform_on_gold_data(
+    dataframe, queue, whiten, num_blocks, dtype
+):
+
+    X = np.array([[-1, -1], [-2, -1], [-3, -2], [1, 1], [2, 1], [3, 2]])
+    X = X.astype(dtype=dtype)
+    incpca = IncrementalPCA(whiten=whiten, batch_size=X.shape[0] // num_blocks)
+
+    X_df = _convert_to_dataframe(X, sycl_queue=queue, target_df=dataframe)
+    transformed_data = incpca.fit_transform(X_df)
+
+    check_pca_on_gold_data(incpca, dtype, whiten, transformed_data)
+
+
+@pytest.mark.parametrize("dataframe,queue", get_dataframes_and_queues())
+@pytest.mark.parametrize("n_components", [None, 1, 5])
+@pytest.mark.parametrize("whiten", [True, False])
+@pytest.mark.parametrize("num_blocks", [1, 10])
+@pytest.mark.parametrize("row_count", [100, 1000])
+@pytest.mark.parametrize("column_count", [10, 100])
+@pytest.mark.parametrize("dtype", [np.float32, np.float64])
+def test_sklearnex_partial_fit_on_random_data(
+    dataframe, queue, n_components, whiten, num_blocks, row_count, column_count, dtype
+):
+    seed = 81
+    gen = np.random.default_rng(seed)
+    X = gen.uniform(low=-0.3, high=+0.7, size=(row_count, column_count))
+    X = X.astype(dtype=dtype)
+    X_split = np.array_split(X, num_blocks)
+    incpca = IncrementalPCA(n_components=n_components, whiten=whiten)
+
+    for i in range(num_blocks):
+        X_split_df = _convert_to_dataframe(
+            X_split[i], sycl_queue=queue, target_df=dataframe
+        )
+        incpca.partial_fit(X_split_df)
+
+    X_df = _convert_to_dataframe(X, sycl_queue=queue, target_df=dataframe)
+    transformed_data = incpca.transform(X_df)
+    check_pca(incpca, dtype, whiten, X, transformed_data)
+
+
+@pytest.mark.parametrize("dataframe,queue", get_dataframes_and_queues())
+@pytest.mark.parametrize("dtype", [np.float32, np.float64])
+def test_sklearnex_incremental_estimatior_pickle(dataframe, queue, dtype):
+    import pickle
+
+    incpca = IncrementalPCA()
+
+    # Check that estimator can be serialized without any data.
+    dump = pickle.dumps(incpca)
+    incpca_loaded = pickle.loads(dump)
+
+    seed = 77
+    gen = np.random.default_rng(seed)
+    X = gen.uniform(low=-0.3, high=+0.7, size=(10, 10))
+    X = X.astype(dtype)
+    X_split = np.array_split(X, 2)
+    X_split_df = _convert_to_dataframe(X_split[0], sycl_queue=queue, target_df=dataframe)
+    incpca.partial_fit(X_split_df)
+    incpca_loaded.partial_fit(X_split_df)
+    dump = pickle.dumps(incpca_loaded)
+    incpca_loaded = pickle.loads(dump)
+    assert incpca.batch_size == incpca_loaded.batch_size
+    assert incpca.n_features_in_ == incpca_loaded.n_features_in_
+    assert incpca.n_samples_seen_ == incpca_loaded.n_samples_seen_
+    if hasattr(incpca, "_parameter_constraints"):
+        assert incpca._parameter_constraints == incpca_loaded._parameter_constraints
+    assert incpca.n_jobs == incpca_loaded.n_jobs
+    X_split_df = _convert_to_dataframe(X_split[1], sycl_queue=queue, target_df=dataframe)
+    incpca.partial_fit(X_split_df)
+    incpca_loaded.partial_fit(X_split_df)
+
+    # Check that estimator can be serialized after partial_fit call.
+    dump = pickle.dumps(incpca)
+    incpca_loaded = pickle.loads(dump)
+
+    assert_allclose(incpca.singular_values_, incpca_loaded.singular_values_, atol=1e-6)
+    assert_allclose(incpca.n_samples_seen_, incpca_loaded.n_samples_seen_, atol=1e-6)
+    assert_allclose(incpca.n_features_in_, incpca_loaded.n_features_in_, atol=1e-6)
+    assert_allclose(incpca.mean_, incpca_loaded.mean_, atol=1e-6)
+    assert_allclose(incpca.var_, incpca_loaded.var_, atol=1e-6)
+    assert_allclose(
+        incpca.explained_variance_, incpca_loaded.explained_variance_, atol=1e-6
+    )
+    assert_allclose(incpca.components_, incpca_loaded.components_, atol=1e-6)
+    assert_allclose(
+        incpca.explained_variance_ratio_,
+        incpca_loaded.explained_variance_ratio_,
+        atol=1e-6,
+    )
+
+    # Check that finalized estimator can be serialized.
+    dump = pickle.dumps(incpca_loaded)
+    incpca_loaded = pickle.loads(dump)
+
+    assert_allclose(incpca.singular_values_, incpca_loaded.singular_values_, atol=1e-6)
+    assert_allclose(incpca.n_samples_seen_, incpca_loaded.n_samples_seen_, atol=1e-6)
+    assert_allclose(incpca.n_features_in_, incpca_loaded.n_features_in_, atol=1e-6)
+    assert_allclose(incpca.mean_, incpca_loaded.mean_, atol=1e-6)
+    assert_allclose(incpca.var_, incpca_loaded.var_, atol=1e-6)
+    assert_allclose(
+        incpca.explained_variance_, incpca_loaded.explained_variance_, atol=1e-6
+    )
+    assert_allclose(incpca.components_, incpca_loaded.components_, atol=1e-6)
+    assert_allclose(
+        incpca.explained_variance_ratio_,
+        incpca_loaded.explained_variance_ratio_,
+        atol=1e-6,
+    )
+
+
+@pytest.mark.parametrize("dataframe,queue", get_dataframes_and_queues("numpy,dpctl"))
+def test_changed_estimated_attributes(with_array_api, dataframe, queue):
+    # check that attributes necessary for the PCA onedal estimator match
+    # changes occurring in the sklearnex estimator
+    X, y = load_iris(return_X_y=True)
+
+    X_0 = _convert_to_dataframe(X[y == 0], sycl_queue=queue, target_df=dataframe)
+    X = _convert_to_dataframe(X, sycl_queue=queue, target_df=dataframe)
+
+    est0 = IncrementalPCA(n_components=4).fit(X_0)
+    est = clone(est0)
+    assert not hasattr(est, "_onedal_estimator")
+    est.fit(X)
+    assert not np.array_equal(_as_numpy(est.transform(X)), _as_numpy(est0.transform(X)))
+
+    # copy over parameters necessary for transform
+    est.mean_ = est0.mean_
+    est.components_ = est0.components_
+    est.explained_variance_ = est0.explained_variance_
+    est.n_components_ = est0.n_components_  # is trivial but exercises the logic
+
+    assert np.array_equal(_as_numpy(est.transform(X)), _as_numpy(est0.transform(X)))
+
+
+@pytest.mark.skipif(
+    not sklearn_check_version("1.5"),
+    reason='svd_solver="auto" does not support sparse inputs',
+)
+@pytest.mark.allow_sklearn_fallback
+def test_create_model_behavior():
+    # verify that fit fallbacks does not break ``transform`` as the oneDAL
+    # model is generated JIT
+
+    X, _ = load_iris(return_X_y=True)
+    # generate a onedal estimator
+    est = IncrementalPCA(n_components=3)
+    X_trans = est.fit_transform(X)
+
+    # force data to sparse for a fallback to sklearn
+    X_sp = sp.csr_matrix(X)
+    est.fit(X_sp)
+    # In the case of a fallback, the model should be set to none by clobbered
+    # fitted attributes
+    assert est._onedal_estimator._onedal_model is None
+
+    X_trans_sparse = est.transform(X)
+    # use allclose as data was fit with sklearn and onedal on the same data
+    # but using different backends
+    assert_allclose(X_trans, X_trans_sparse)

scikit_learn_intelex-2025.10.0.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-2025.10.0.data/data/Lib/site-packages/sklearnex/spmd/basic_statistics/basic_statistics.py

@@ -0,0 +1,24 @@
+# ==============================================================================
+# 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 onedal.spmd.basic_statistics import BasicStatistics as onedal_BasicStatistics
+
+from ...basic_statistics import BasicStatistics as BasicStatistics_Batch
+
+
+class BasicStatistics(BasicStatistics_Batch):
+    __doc__ = BasicStatistics_Batch.__doc__
+    _onedal_basic_statistics = staticmethod(onedal_BasicStatistics)

scikit_learn_intelex-2025.10.0.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
+    )

scikit_learn_intelex-2025.10.0.data/data/Lib/site-packages/sklearnex/spmd/basic_statistics/tests/test_basic_statistics_spmd.py

@@ -0,0 +1,117 @@
+# ==============================================================================
+# 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.utils import options_and_tests
+from onedal.tests.utils._dataframes_support import (
+    _as_numpy,
+    _convert_to_dataframe,
+    get_dataframes_and_queues,
+)
+from sklearnex import config_context
+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 = BasicStatistics_SPMD()
+    spmd_result = spmd.fit(local_dpt_data)
+    batch_result = BasicStatistics_Batch().fit(data)
+
+    for option in options_and_tests:
+        attr = option + "_"
+        assert_allclose(getattr(spmd_result, attr), getattr(batch_result, attr))
+
+
+@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.parametrize("use_raw_input", [True, False])
+@pytest.mark.mpi
+def test_basic_stats_spmd_synthetic(
+    n_samples, n_features, dataframe, queue, dtype, use_raw_input
+):
+    # 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
+    # Configure raw input status for spmd estimator
+    with config_context(use_raw_input=use_raw_input):
+        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 options_and_tests:
+        attr = option + "_"
+        assert_allclose(
+            _as_numpy(getattr(spmd_result, attr)),
+            _as_numpy(getattr(batch_result, attr)),
+            atol=tol,
+            rtol=tol,
+        )