scikit-learn-intelex 2024.1.0__py312-none-manylinux1_x86_64.whl → 2024.3.0__py312-none-manylinux1_x86_64.whl

sklearnex/ensemble/_forest.py

@@ -42,11 +42,10 @@ from sklearn.utils.validation import (
     check_X_y,
 )
 
+from daal4py.sklearn._n_jobs_support import control_n_jobs
 from daal4py.sklearn._utils import (
     check_tree_nodes,
-    control_n_jobs,
     daal_check_version,
-    run_with_n_jobs,
     sklearn_check_version,
 )
 from onedal.ensemble import ExtraTreesClassifier as onedal_ExtraTreesClassifier
@@ -78,7 +77,6 @@ if sklearn_check_version("1.4"):
 class BaseForest(ABC):
     _onedal_factory = None
 
-    @run_with_n_jobs
     def _onedal_fit(self, X, y, sample_weight=None, queue=None):
         if sklearn_check_version("0.24"):
             X, y = self._validate_data(
@@ -455,14 +453,12 @@ class ForestClassifier(sklearn_ForestClassifier, BaseForest):
 
         # The estimator is checked against the class attribute for conformance.
         # This should only trigger if the user uses this class directly.
-        if (
-            self.estimator.__class__ == DecisionTreeClassifier
-            and self._onedal_factory != onedal_RandomForestClassifier
+        if self.estimator.__class__ == DecisionTreeClassifier and not issubclass(
+            self._onedal_factory, onedal_RandomForestClassifier
         ):
             self._onedal_factory = onedal_RandomForestClassifier
-        elif (
-            self.estimator.__class__ == ExtraTreeClassifier
-            and self._onedal_factory != onedal_ExtraTreesClassifier
+        elif self.estimator.__class__ == ExtraTreeClassifier and not issubclass(
+            self._onedal_factory, onedal_ExtraTreesClassifier
         ):
             self._onedal_factory = onedal_ExtraTreesClassifier
 
@@ -749,7 +745,7 @@ class ForestClassifier(sklearn_ForestClassifier, BaseForest):
                         or self.estimator.__class__ == DecisionTreeClassifier,
                         "ExtraTrees only supported starting from oneDAL version 2023.1",
                     ),
-                    (sample_weight is not None, "sample_weight is not supported."),
+                    (sample_weight is None, "sample_weight is not supported."),
                 ]
             )
 
@@ -787,7 +783,6 @@ class ForestClassifier(sklearn_ForestClassifier, BaseForest):
 
         return patching_status
 
-    @run_with_n_jobs
     def _onedal_predict(self, X, queue=None):
         X = check_array(
             X,
@@ -802,7 +797,6 @@ class ForestClassifier(sklearn_ForestClassifier, BaseForest):
         res = self._onedal_estimator.predict(X, queue=queue)
         return np.take(self.classes_, res.ravel().astype(np.int64, casting="unsafe"))
 
-    @run_with_n_jobs
     def _onedal_predict_proba(self, X, queue=None):
         X = check_array(X, dtype=[np.float64, np.float32], force_all_finite=False)
         check_is_fitted(self, "_onedal_estimator")
@@ -847,14 +841,12 @@ class ForestRegressor(sklearn_ForestRegressor, BaseForest):
 
         # The splitter is checked against the class attribute for conformance
         # This should only trigger if the user uses this class directly.
-        if (
-            self.estimator.__class__ == DecisionTreeRegressor
-            and self._onedal_factory != onedal_RandomForestRegressor
+        if self.estimator.__class__ == DecisionTreeRegressor and not issubclass(
+            self._onedal_factory, onedal_RandomForestRegressor
         ):
             self._onedal_factory = onedal_RandomForestRegressor
-        elif (
-            self.estimator.__class__ == ExtraTreeRegressor
-            and self._onedal_factory != onedal_ExtraTreesRegressor
+        elif self.estimator.__class__ == ExtraTreeRegressor and not issubclass(
+            self._onedal_factory, onedal_ExtraTreesRegressor
         ):
             self._onedal_factory = onedal_ExtraTreesRegressor
 
@@ -1060,7 +1052,7 @@ class ForestRegressor(sklearn_ForestRegressor, BaseForest):
                         or self.estimator.__class__ == DecisionTreeClassifier,
                         "ExtraTrees only supported starting from oneDAL version 2023.1",
                     ),
-                    (sample_weight is not None, "sample_weight is not supported."),
+                    (sample_weight is None, "sample_weight is not supported."),
                 ]
             )
 
@@ -1096,7 +1088,6 @@ class ForestRegressor(sklearn_ForestRegressor, BaseForest):
 
         return patching_status
 
-    @run_with_n_jobs
     def _onedal_predict(self, X, queue=None):
         X = check_array(
             X, dtype=[np.float64, np.float32], force_all_finite=False
@@ -1138,7 +1129,7 @@ class ForestRegressor(sklearn_ForestRegressor, BaseForest):
     predict.__doc__ = sklearn_ForestRegressor.predict.__doc__
 
 
-@control_n_jobs
+@control_n_jobs(decorated_methods=["fit", "predict", "predict_proba"])
 class RandomForestClassifier(ForestClassifier):
     __doc__ = sklearn_RandomForestClassifier.__doc__
     _onedal_factory = onedal_RandomForestClassifier
@@ -1348,7 +1339,7 @@ class RandomForestClassifier(ForestClassifier):
             self.min_bin_size = min_bin_size
 
 
-@control_n_jobs
+@control_n_jobs(decorated_methods=["fit", "predict"])
 class RandomForestRegressor(ForestRegressor):
     __doc__ = sklearn_RandomForestRegressor.__doc__
     _onedal_factory = onedal_RandomForestRegressor
@@ -1549,7 +1540,7 @@ class RandomForestRegressor(ForestRegressor):
             self.min_bin_size = min_bin_size
 
 
-@control_n_jobs
+@control_n_jobs(decorated_methods=["fit", "predict", "predict_proba"])
 class ExtraTreesClassifier(ForestClassifier):
     __doc__ = sklearn_ExtraTreesClassifier.__doc__
     _onedal_factory = onedal_ExtraTreesClassifier
@@ -1759,7 +1750,7 @@ class ExtraTreesClassifier(ForestClassifier):
             self.min_bin_size = min_bin_size
 
 
-@control_n_jobs
+@control_n_jobs(decorated_methods=["fit", "predict"])
 class ExtraTreesRegressor(ForestRegressor):
     __doc__ = sklearn_ExtraTreesRegressor.__doc__
     _onedal_factory = onedal_ExtraTreesRegressor

sklearnex/ensemble/tests/test_forest.py

@@ -45,11 +45,7 @@ def test_sklearnex_import_rf_classifier(dataframe, queue):
     assert_allclose([1], _as_numpy(rf.predict([[0, 0, 0, 0]])))
 
 
-# TODO:
-# investigate failure for `dpnp.ndarrays` and `dpctl.tensors` on `GPU`
-@pytest.mark.parametrize(
-    "dataframe,queue", get_dataframes_and_queues(device_filter_="cpu")
-)
+@pytest.mark.parametrize("dataframe,queue", get_dataframes_and_queues())
 def test_sklearnex_import_rf_regression(dataframe, queue):
     from sklearnex.ensemble import RandomForestRegressor
 
@@ -59,17 +55,17 @@ def test_sklearnex_import_rf_regression(dataframe, queue):
     rf = RandomForestRegressor(max_depth=2, random_state=0).fit(X, y)
     assert "sklearnex" in rf.__module__
     pred = _as_numpy(rf.predict([[0, 0, 0, 0]]))
-    if daal_check_version((2024, "P", 0)):
-        assert_allclose([-6.971], pred, atol=1e-2)
+
+    if queue is not None and queue.sycl_device.is_gpu:
+        assert_allclose([-0.011208], pred, atol=1e-2)
     else:
-        assert_allclose([-6.839], pred, atol=1e-2)
+        if daal_check_version((2024, "P", 0)):
+            assert_allclose([-6.971], pred, atol=1e-2)
+        else:
+            assert_allclose([-6.839], pred, atol=1e-2)
 
 
-# TODO:
-# investigate failure for `dpnp.ndarrays` and `dpctl.tensors` on `GPU`
-@pytest.mark.parametrize(
-    "dataframe,queue", get_dataframes_and_queues(device_filter_="cpu")
-)
+@pytest.mark.parametrize("dataframe,queue", get_dataframes_and_queues())
 def test_sklearnex_import_et_classifier(dataframe, queue):
     from sklearnex.ensemble import ExtraTreesClassifier
 
@@ -90,11 +86,7 @@ def test_sklearnex_import_et_classifier(dataframe, queue):
     assert_allclose([1], _as_numpy(rf.predict([[0, 0, 0, 0]])))
 
 
-# TODO:
-# investigate failure for `dpnp.ndarrays` and `dpctl.tensors` on `GPU`
-@pytest.mark.parametrize(
-    "dataframe,queue", get_dataframes_and_queues(device_filter_="cpu")
-)
+@pytest.mark.parametrize("dataframe,queue", get_dataframes_and_queues())
 def test_sklearnex_import_et_regression(dataframe, queue):
     from sklearnex.ensemble import ExtraTreesRegressor
 
@@ -114,4 +106,8 @@ def test_sklearnex_import_et_regression(dataframe, queue):
             ]
         )
     )
-    assert_allclose([0.445], pred, atol=1e-2)
+
+    if queue is not None and queue.sycl_device.is_gpu:
+        assert_allclose([1.909769], pred, atol=1e-2)
+    else:
+        assert_allclose([0.445], pred, atol=1e-2)

sklearnex/linear_model/__init__.py

@@ -16,14 +16,13 @@
 
 from .coordinate_descent import ElasticNet, Lasso
 from .linear import LinearRegression
-from .logistic_path import LogisticRegression, logistic_regression_path
+from .logistic_regression import LogisticRegression
 from .ridge import Ridge
 
 __all__ = [
     "Ridge",
     "LinearRegression",
     "LogisticRegression",
-    "logistic_regression_path",
     "ElasticNet",
     "Lasso",
 ]

sklearnex/linear_model/linear.py

@@ -65,13 +65,8 @@ if daal_check_version((2023, "P", 100)):
     import numpy as np
     from sklearn.linear_model import LinearRegression as sklearn_LinearRegression
 
-    from daal4py.sklearn._utils import (
-        control_n_jobs,
-        get_dtype,
-        make2d,
-        run_with_n_jobs,
-        sklearn_check_version,
-    )
+    from daal4py.sklearn._n_jobs_support import control_n_jobs
+    from daal4py.sklearn._utils import get_dtype, make2d, sklearn_check_version
 
     from .._device_offload import dispatch, wrap_output_data
     from .._utils import (
@@ -93,7 +88,7 @@ if daal_check_version((2023, "P", 100)):
     from onedal.utils import _num_features, _num_samples
 
     @register_hyperparameters({"fit": get_hyperparameters("linear_regression", "train")})
-    @control_n_jobs
+    @control_n_jobs(decorated_methods=["fit", "predict"])
     class LinearRegression(sklearn_LinearRegression, BaseLinearRegression):
         __doc__ = sklearn_LinearRegression.__doc__
         intercept_, coef_ = None, None
@@ -330,7 +325,6 @@ if daal_check_version((2023, "P", 100)):
             onedal_params = {"fit_intercept": self.fit_intercept, "copy_X": self.copy_X}
             self._onedal_estimator = onedal_LinearRegression(**onedal_params)
 
-        @run_with_n_jobs
         def _onedal_fit(self, X, y, sample_weight, queue=None):
             assert sample_weight is None
 
@@ -369,7 +363,6 @@ if daal_check_version((2023, "P", 100)):
                 del self._onedal_estimator
                 super().fit(X, y)
 
-        @run_with_n_jobs
         def _onedal_predict(self, X, queue=None):
             X = self._validate_data(X, accept_sparse=False, reset=False)
             if not hasattr(self, "_onedal_estimator"):

sklearnex/{preview/linear_model → linear_model}/logistic_regression.py

@@ -1,5 +1,5 @@
 # ===============================================================================
-# Copyright 2023 Intel Corporation
+# 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.
@@ -17,14 +17,11 @@
 import logging
 from abc import ABC
 
-import sklearn.linear_model._logistic as logistic_module
-
 from daal4py.sklearn._utils import daal_check_version
 from daal4py.sklearn.linear_model.logistic_path import (
-    LogisticRegression,
-    daal4py_predict,
-    logistic_regression_path,
+    LogisticRegression as LogisticRegression_daal4py,
 )
+from daal4py.sklearn.linear_model.logistic_path import daal4py_fit, daal4py_predict
 
 
 class BaseLogisticRegression(ABC):
@@ -43,14 +40,18 @@ if daal_check_version((2024, "P", 1)):
     from sklearn.linear_model import LogisticRegression as sklearn_LogisticRegression
     from sklearn.utils.validation import check_X_y
 
+    from daal4py.sklearn._n_jobs_support import control_n_jobs
     from daal4py.sklearn._utils import sklearn_check_version
     from onedal.linear_model import LogisticRegression as onedal_LogisticRegression
     from onedal.utils import _num_features, _num_samples
 
-    from ..._device_offload import dispatch, wrap_output_data
-    from ..._utils import PatchingConditionsChain, get_patch_message
-    from ...utils.validation import _assert_all_finite
+    from .._device_offload import dispatch, wrap_output_data
+    from .._utils import PatchingConditionsChain, get_patch_message
+    from ..utils.validation import _assert_all_finite
 
+    @control_n_jobs(
+        decorated_methods=["fit", "predict", "predict_proba", "predict_log_proba"]
+    )
     class LogisticRegression(sklearn_LogisticRegression, BaseLogisticRegression):
         __doc__ = sklearn_LogisticRegression.__doc__
         intercept_, coef_, n_iter_ = None, None, None
@@ -97,6 +98,8 @@ if daal_check_version((2024, "P", 1)):
                 l1_ratio=l1_ratio,
             )
 
+        _onedal_cpu_fit = daal4py_fit
+
         def fit(self, X, y, sample_weight=None):
             if sklearn_check_version("1.0"):
                 self._check_feature_names(X, reset=True)
@@ -160,10 +163,8 @@ if daal_check_version((2024, "P", 1)):
         def _test_type_and_finiteness(self, X_in):
             X = np.asarray(X_in)
 
-            dtype = X.dtype
-            if "complex" in str(type(dtype)):
+            if np.iscomplexobj(X):
                 return False
-
             try:
                 _assert_all_finite(X)
             except BaseException:
@@ -184,7 +185,10 @@ if daal_check_version((2024, "P", 1)):
                 [
                     (self.penalty == "l2", "Only l2 penalty is supported."),
                     (self.dual == False, "dual=True is not supported."),
-                    (self.intercept_scaling == 1, "Intercept scaling is not supported."),
+                    (
+                        self.intercept_scaling == 1,
+                        "Intercept scaling is not supported.",
+                    ),
                     (self.class_weight is None, "Class weight is not supported"),
                     (self.solver == "newton-cg", "Only newton-cg solver is supported."),
                     (
@@ -229,7 +233,10 @@ if daal_check_version((2024, "P", 1)):
                     (n_samples > 0, "Number of samples is less than 1."),
                     (not issparse(*data), "Sparse input is not supported."),
                     (not model_is_sparse, "Sparse coefficients are not supported."),
-                    (hasattr(self, "_onedal_estimator"), "oneDAL model was not trained."),
+                    (
+                        hasattr(self, "_onedal_estimator"),
+                        "oneDAL model was not trained.",
+                    ),
                 ]
             )
             if not dal_ready:
@@ -268,15 +275,6 @@ if daal_check_version((2024, "P", 1)):
             }
             self._onedal_estimator = onedal_LogisticRegression(**onedal_params)
 
-        def _onedal_cpu_fit(self, X, y, sample_weight):
-            which, what = logistic_module, "_logistic_regression_path"
-            replacer = logistic_regression_path
-            descriptor = getattr(which, what, None)
-            setattr(which, what, replacer)
-            clf = super().fit(X, y, sample_weight)
-            setattr(which, what, descriptor)
-            return clf
-
         def _onedal_fit(self, X, y, sample_weight, queue=None):
             if queue is None or queue.sycl_device.is_cpu:
                 return self._onedal_cpu_fit(X, y, sample_weight)
@@ -313,38 +311,32 @@ if daal_check_version((2024, "P", 1)):
                 return daal4py_predict(self, X, "computeClassLabels")
 
             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_
-                self._onedal_estimator.classes_ = self.classes_
-
+            assert hasattr(self, "_onedal_estimator")
             return self._onedal_estimator.predict(X, queue=queue)
 
         def _onedal_predict_proba(self, X, queue=None):
             if queue is None or queue.sycl_device.is_cpu:
                 return daal4py_predict(self, X, "computeClassProbabilities")
-            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_
 
+            X = self._validate_data(X, accept_sparse=False, reset=False)
+            assert hasattr(self, "_onedal_estimator")
             return self._onedal_estimator.predict_proba(X, queue=queue)
 
         def _onedal_predict_log_proba(self, X, queue=None):
             if queue is None or queue.sycl_device.is_cpu:
                 return daal4py_predict(self, X, "computeClassLogProbabilities")
-            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_
 
+            X = self._validate_data(X, accept_sparse=False, reset=False)
+            assert hasattr(self, "_onedal_estimator")
             return self._onedal_estimator.predict_log_proba(X, queue=queue)
 
+        fit.__doc__ = sklearn_LogisticRegression.fit.__doc__
+        predict.__doc__ = sklearn_LogisticRegression.predict.__doc__
+        predict_proba.__doc__ = sklearn_LogisticRegression.predict_proba.__doc__
+        predict_log_proba.__doc__ = sklearn_LogisticRegression.predict_log_proba.__doc__
+
 else:
-    from daal4py.sklearn.linear_model import LogisticRegression
+    LogisticRegression = LogisticRegression_daal4py
 
     logging.warning(
         "Sklearnex LogisticRegression requires oneDAL version >= 2024.0.1 "

sklearnex/linear_model/tests/test_logreg.py

@@ -14,15 +14,78 @@
 # limitations under the License.
 # ===============================================================================
 
-import numpy as np
-from numpy.testing import assert_allclose
-from sklearn.datasets import load_iris
+import pytest
+from sklearn.datasets import load_breast_cancer, load_iris
+from sklearn.metrics import accuracy_score
+from sklearn.model_selection import train_test_split
 
+from daal4py.sklearn._utils import daal_check_version
+from onedal.tests.utils._dataframes_support import (
+    _as_numpy,
+    _convert_to_dataframe,
+    get_dataframes_and_queues,
+)
 
-def test_sklearnex_import():
+
+def prepare_input(X, y, dataframe, queue):
+    X_train, X_test, y_train, y_test = train_test_split(
+        X, y, train_size=0.8, random_state=42
+    )
+    X_train = _convert_to_dataframe(X_train, sycl_queue=queue, target_df=dataframe)
+    y_train = _convert_to_dataframe(y_train, sycl_queue=queue, target_df=dataframe)
+    X_test = _convert_to_dataframe(X_test, sycl_queue=queue, target_df=dataframe)
+    return X_train, X_test, y_train, y_test
+
+
+@pytest.mark.parametrize(
+    "dataframe,queue",
+    get_dataframes_and_queues(device_filter_="cpu"),
+)
+def test_sklearnex_multiclass_classification(dataframe, queue):
     from sklearnex.linear_model import LogisticRegression
 
     X, y = load_iris(return_X_y=True)
-    logreg = LogisticRegression(random_state=0, max_iter=200).fit(X, y)
-    assert "daal4py" in logreg.__module__
-    assert_allclose(logreg.score(X, y), 0.9733, atol=1e-3)
+    X_train, X_test, y_train, y_test = prepare_input(X, y, dataframe, queue)
+
+    logreg = LogisticRegression(fit_intercept=True, solver="lbfgs", max_iter=200).fit(
+        X_train, y_train
+    )
+
+    if daal_check_version((2024, "P", 1)):
+        assert "sklearnex" in logreg.__module__
+    else:
+        assert "daal4py" in logreg.__module__
+
+    y_pred = _as_numpy(logreg.predict(X_test))
+    assert accuracy_score(y_test, y_pred) > 0.99
+
+
+@pytest.mark.parametrize(
+    "dataframe,queue",
+    get_dataframes_and_queues(),
+)
+def test_sklearnex_binary_classification(dataframe, queue):
+    from sklearnex.linear_model import LogisticRegression
+
+    X, y = load_breast_cancer(return_X_y=True)
+    X_train, X_test, y_train, y_test = prepare_input(X, y, dataframe, queue)
+
+    logreg = LogisticRegression(fit_intercept=True, solver="newton-cg", max_iter=100).fit(
+        X_train, y_train
+    )
+
+    if daal_check_version((2024, "P", 1)):
+        assert "sklearnex" in logreg.__module__
+    else:
+        assert "daal4py" in logreg.__module__
+    if (
+        dataframe != "numpy"
+        and queue is not None
+        and queue.sycl_device.is_gpu
+        and daal_check_version((2024, "P", 1))
+    ):
+        # fit was done on gpu
+        assert hasattr(logreg, "_onedal_estimator")
+
+    y_pred = _as_numpy(logreg.predict(X_test))
+    assert accuracy_score(y_test, y_pred) > 0.95

sklearnex/neighbors/__init__.py

@@ -14,10 +14,10 @@
 # limitations under the License.
 # ===============================================================================
 
+from ._lof import LocalOutlierFactor
 from .knn_classification import KNeighborsClassifier
 from .knn_regression import KNeighborsRegressor
 from .knn_unsupervised import NearestNeighbors
-from .lof import LocalOutlierFactor
 
 __all__ = [
     "KNeighborsClassifier",