scikit-learn-intelex 2024.1.0__py311-none-manylinux1_x86_64.whl → 2024.4.0__py311-none-manylinux1_x86_64.whl

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):
@@ -41,16 +38,28 @@ if daal_check_version((2024, "P", 1)):
     import numpy as np
     from scipy.sparse import issparse
     from sklearn.linear_model import LogisticRegression as sklearn_LogisticRegression
+    from sklearn.metrics import accuracy_score
+    from sklearn.utils.multiclass import type_of_target
     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 onedal.utils import _num_samples
+
+    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",
+            "score",
+        ]
+    )
     class LogisticRegression(sklearn_LogisticRegression, BaseLogisticRegression):
         __doc__ = sklearn_LogisticRegression.__doc__
         intercept_, coef_, n_iter_ = None, None, None
@@ -71,9 +80,9 @@ if daal_check_version((2024, "P", 1)):
             intercept_scaling=1,
             class_weight=None,
             random_state=None,
-            solver="lbfgs" if sklearn_check_version("0.22") else "liblinear",
+            solver="lbfgs",
             max_iter=100,
-            multi_class="auto" if sklearn_check_version("0.22") else "ovr",
+            multi_class="auto",
             verbose=0,
             warm_start=False,
             n_jobs=None,
@@ -97,6 +106,8 @@ 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)
@@ -157,13 +168,32 @@ if daal_check_version((2024, "P", 1)):
                 X,
             )
 
+        @wrap_output_data
+        def score(self, X, y, sample_weight=None):
+            if sklearn_check_version("1.0"):
+                self._check_feature_names(X, reset=False)
+            return dispatch(
+                self,
+                "score",
+                {
+                    "onedal": self.__class__._onedal_score,
+                    "sklearn": sklearn_LogisticRegression.score,
+                },
+                X,
+                y,
+                sample_weight=sample_weight,
+            )
+
+        def _onedal_score(self, X, y, sample_weight=None, queue=None):
+            return accuracy_score(
+                y, self._onedal_predict(X, queue=queue), sample_weight=sample_weight
+            )
+
         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 +214,10 @@ 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."),
                     (
@@ -194,6 +227,10 @@ if daal_check_version((2024, "P", 1)):
                     (self.warm_start == False, "Warm start is not supported."),
                     (self.l1_ratio is None, "l1 ratio is not supported."),
                     (sample_weight is None, "Sample weight is not supported."),
+                    (
+                        type_of_target(y) == "binary",
+                        "Only binary classification is supported",
+                    ),
                 ]
             )
 
@@ -212,24 +249,34 @@ if daal_check_version((2024, "P", 1)):
             return patching_status
 
         def _onedal_gpu_predict_supported(self, method_name, *data):
-            assert method_name in ["predict", "predict_proba", "predict_log_proba"]
-            assert len(data) == 1
+            assert method_name in [
+                "predict",
+                "predict_proba",
+                "predict_log_proba",
+                "score",
+            ]
 
             class_name = self.__class__.__name__
             patching_status = PatchingConditionsChain(
                 f"sklearn.linear_model.{class_name}.{method_name}"
             )
 
-            n_samples = _num_samples(*data)
+            n_samples = _num_samples(data[0])
             model_is_sparse = issparse(self.coef_) or (
                 self.fit_intercept and issparse(self.intercept_)
             )
             dal_ready = patching_status.and_conditions(
                 [
                     (n_samples > 0, "Number of samples is less than 1."),
-                    (not issparse(*data), "Sparse input is not supported."),
+                    (
+                        not any([issparse(i) for i in 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:
@@ -244,7 +291,7 @@ if daal_check_version((2024, "P", 1)):
         def _onedal_gpu_supported(self, method_name, *data):
             if method_name == "fit":
                 return self._onedal_gpu_fit_supported(method_name, *data)
-            if method_name in ["predict", "predict_proba", "predict_log_proba"]:
+            if method_name in ["predict", "predict_proba", "predict_log_proba", "score"]:
                 return self._onedal_gpu_predict_supported(method_name, *data)
             raise RuntimeError(
                 f"Unknown method {method_name} in {self.__class__.__name__}"
@@ -268,15 +315,6 @@ 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 +351,33 @@ 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__
+        score.__doc__ = sklearn_LogisticRegression.score.__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_linear.py

@@ -28,26 +28,33 @@ from onedal.tests.utils._dataframes_support import (
 
 
 @pytest.mark.parametrize("dataframe,queue", get_dataframes_and_queues())
+@pytest.mark.parametrize("dtype", [np.float32, np.float64])
 @pytest.mark.parametrize("macro_block", [None, 1024])
-def test_sklearnex_import_linear(dataframe, queue, macro_block):
+def test_sklearnex_import_linear(dataframe, queue, dtype, macro_block):
     from sklearnex.linear_model import LinearRegression
 
     X = np.array([[1, 1], [1, 2], [2, 2], [2, 3]])
     y = np.dot(X, np.array([1, 2])) + 3
+    X = X.astype(dtype=dtype)
+    y = y.astype(dtype=dtype)
     X = _convert_to_dataframe(X, sycl_queue=queue, target_df=dataframe)
     y = _convert_to_dataframe(y, sycl_queue=queue, target_df=dataframe)
+
     linreg = LinearRegression()
     if daal_check_version((2024, "P", 0)) and macro_block is not None:
         hparams = linreg.get_hyperparameters("fit")
         hparams.cpu_macro_block = macro_block
         hparams.gpu_macro_block = macro_block
+
     linreg.fit(X, y)
-    if daal_check_version((2023, "P", 100)):
-        assert hasattr(linreg, "_onedal_estimator")
+
+    assert hasattr(linreg, "_onedal_estimator")
     assert "sklearnex" in linreg.__module__
     assert linreg.n_features_in_ == 2
-    assert_allclose(_as_numpy(linreg.intercept_), 3.0)
-    assert_allclose(_as_numpy(linreg.coef_), [1.0, 2.0])
+
+    tol = 1e-5 if X.dtype == np.float32 else 1e-7
+    assert_allclose(_as_numpy(linreg.intercept_), 3.0, rtol=tol)
+    assert_allclose(_as_numpy(linreg.coef_), [1.0, 2.0], rtol=tol)
 
 
 def test_sklearnex_import_ridge():
@@ -80,3 +87,31 @@ def test_sklearnex_import_elastic():
     assert "daal4py" in elasticnet.__module__
     assert_allclose(elasticnet.intercept_, 1.451, atol=1e-3)
     assert_allclose(elasticnet.coef_, [18.838, 64.559], atol=1e-3)
+
+
+@pytest.mark.parametrize("dataframe,queue", get_dataframes_and_queues())
+@pytest.mark.parametrize("dtype", [np.float32, np.float64])
+def test_sklearnex_reconstruct_model(dataframe, queue, dtype):
+    from sklearnex.linear_model import LinearRegression
+
+    seed = 42
+    num_samples = 3500
+    num_features, num_targets = 14, 9
+
+    gen = np.random.default_rng(seed)
+    intercept = gen.random(size=num_targets, dtype=dtype)
+    coef = gen.random(size=(num_targets, num_features), dtype=dtype).T
+
+    X = gen.random(size=(num_samples, num_features), dtype=dtype)
+    gtr = X @ coef + intercept[np.newaxis, :]
+
+    X = _convert_to_dataframe(X, sycl_queue=queue, target_df=dataframe)
+
+    linreg = LinearRegression(fit_intercept=True)
+    linreg.coef_ = coef.T
+    linreg.intercept_ = intercept
+
+    y_pred = linreg.predict(X)
+
+    tol = 1e-5 if X.dtype == np.float32 else 1e-7
+    assert_allclose(gtr, _as_numpy(y_pred), rtol=tol)

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",

sklearnex/neighbors/_lof.py

@@ -0,0 +1,221 @@
+# ===============================================================================
+# 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 warnings
+
+import numpy as np
+from sklearn.neighbors import LocalOutlierFactor as sklearn_LocalOutlierFactor
+from sklearn.utils.metaestimators import available_if
+from sklearn.utils.validation import check_is_fitted
+
+from daal4py.sklearn._n_jobs_support import control_n_jobs
+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
+
+
+@control_n_jobs(decorated_methods=["fit", "_kneighbors"])
+class LocalOutlierFactor(KNeighborsDispatchingBase, sklearn_LocalOutlierFactor):
+    __doc__ = (
+        sklearn_LocalOutlierFactor.__doc__
+        + "\n NOTE: When X=None, methods kneighbors, kneighbors_graph, and predict will"
+        + "\n only output numpy arrays. In that case, the only way to offload to gpu"
+        + "\n is to use a global queue (e.g. using config_context)"
+    )
+    if sklearn_check_version("1.2"):
+        _parameter_constraints: dict = {
+            **sklearn_LocalOutlierFactor._parameter_constraints
+        }
+
+    # Only certain methods should be taken from knn to prevent code
+    # duplication. Inheriting would yield a complicated inheritance
+    # structure and violate the sklearn inheritance path.
+    _save_attributes = NearestNeighbors._save_attributes
+    _onedal_knn_fit = NearestNeighbors._onedal_fit
+    _onedal_kneighbors = NearestNeighbors._onedal_kneighbors
+
+    def _onedal_fit(self, X, y, queue=None):
+        if sklearn_check_version("1.2"):
+            self._validate_params()
+
+        self._onedal_knn_fit(X, y, queue)
+
+        if self.contamination != "auto":
+            if not (0.0 < self.contamination <= 0.5):
+                raise ValueError(
+                    "contamination must be in (0, 0.5], " "got: %f" % self.contamination
+                )
+
+        n_samples = self.n_samples_fit_
+
+        if self.n_neighbors > n_samples:
+            warnings.warn(
+                "n_neighbors (%s) is greater than the "
+                "total number of samples (%s). n_neighbors "
+                "will be set to (n_samples - 1) for estimation."
+                % (self.n_neighbors, n_samples)
+            )
+        self.n_neighbors_ = max(1, min(self.n_neighbors, n_samples - 1))
+
+        (
+            self._distances_fit_X_,
+            _neighbors_indices_fit_X_,
+        ) = self._onedal_kneighbors(n_neighbors=self.n_neighbors_, queue=queue)
+
+        # Sklearn includes a check for float32 at this point which may not be
+        # necessary for onedal
+
+        self._lrd = self._local_reachability_density(
+            self._distances_fit_X_, _neighbors_indices_fit_X_
+        )
+
+        # Compute lof score over training samples to define offset_:
+        lrd_ratios_array = self._lrd[_neighbors_indices_fit_X_] / self._lrd[:, np.newaxis]
+
+        self.negative_outlier_factor_ = -np.mean(lrd_ratios_array, axis=1)
+
+        if self.contamination == "auto":
+            # inliers score around -1 (the higher, the less abnormal).
+            self.offset_ = -1.5
+        else:
+            self.offset_ = np.percentile(
+                self.negative_outlier_factor_, 100.0 * self.contamination
+            )
+
+        return self
+
+    def fit(self, X, y=None):
+        result = dispatch(
+            self,
+            "fit",
+            {
+                "onedal": self.__class__._onedal_fit,
+                "sklearn": sklearn_LocalOutlierFactor.fit,
+            },
+            X,
+            None,
+        )
+        return result
+
+    def _predict(self, X=None):
+        check_is_fitted(self)
+
+        if X is not None:
+            xp, _ = get_namespace(X)
+            output = self.decision_function(X) < 0
+            is_inlier = xp.ones_like(output, dtype=int)
+            is_inlier[output] = -1
+        else:
+            is_inlier = np.ones(self.n_samples_fit_, dtype=int)
+            is_inlier[self.negative_outlier_factor_ < self.offset_] = -1
+
+        return is_inlier
+
+    # This had to be done because predict loses the queue when no
+    # argument is given and it is a dpctl tensor or dpnp array.
+    # This would cause issues in fit_predict. Also, available_if
+    # is hard to unwrap, and this is the most straighforward way.
+    @available_if(sklearn_LocalOutlierFactor._check_novelty_fit_predict)
+    @wrap_output_data
+    def fit_predict(self, X, y=None):
+        """Fit the model to the training set X and return the labels.
+
+        **Not available for novelty detection (when novelty is set to True).**
+        Label is 1 for an inlier and -1 for an outlier according to the LOF
+        score and the contamination parameter.
+
+        Parameters
+        ----------
+        X : {array-like, sparse matrix} of shape (n_samples, n_features), default=None
+            The query sample or samples to compute the Local Outlier Factor
+            w.r.t. the training samples.
+
+        y : Ignored
+            Not used, present for API consistency by convention.
+
+        Returns
+        -------
+        is_inlier : ndarray of shape (n_samples,)
+            Returns -1 for anomalies/outliers and 1 for inliers.
+        """
+        return self.fit(X)._predict()
+
+    def _kneighbors(self, X=None, n_neighbors=None, return_distance=True):
+        check_is_fitted(self)
+        if sklearn_check_version("1.0") and X is not None:
+            self._check_feature_names(X, reset=False)
+        return dispatch(
+            self,
+            "kneighbors",
+            {
+                "onedal": self.__class__._onedal_kneighbors,
+                "sklearn": sklearn_LocalOutlierFactor.kneighbors,
+            },
+            X,
+            n_neighbors=n_neighbors,
+            return_distance=return_distance,
+        )
+
+    kneighbors = wrap_output_data(_kneighbors)
+
+    @available_if(sklearn_LocalOutlierFactor._check_novelty_score_samples)
+    @wrap_output_data
+    def score_samples(self, X):
+        """Opposite of the Local Outlier Factor of X.
+
+        It is the opposite as bigger is better, i.e. large values correspond
+        to inliers.
+
+        **Only available for novelty detection (when novelty is set to True).**
+        The argument X is supposed to contain *new data*: if X contains a
+        point from training, it considers the later in its own neighborhood.
+        Also, the samples in X are not considered in the neighborhood of any
+        point. Because of this, the scores obtained via ``score_samples`` may
+        differ from the standard LOF scores.
+        The standard LOF scores for the training data is available via the
+        ``negative_outlier_factor_`` attribute.
+
+        Parameters
+        ----------
+        X : {array-like, sparse matrix} of shape (n_samples, n_features)
+            The query sample or samples to compute the Local Outlier Factor
+            w.r.t. the training samples.
+
+        Returns
+        -------
+        opposite_lof_scores : ndarray of shape (n_samples,)
+            The opposite of the Local Outlier Factor of each input samples.
+            The lower, the more abnormal.
+        """
+        check_is_fitted(self)
+
+        distances_X, neighbors_indices_X = self._kneighbors(
+            X, n_neighbors=self.n_neighbors_
+        )
+
+        X_lrd = self._local_reachability_density(
+            distances_X,
+            neighbors_indices_X,
+        )
+
+        lrd_ratios_array = self._lrd[neighbors_indices_X] / X_lrd[:, np.newaxis]
+
+        return -np.mean(lrd_ratios_array, axis=1)
+
+    fit.__doc__ = sklearn_LocalOutlierFactor.fit.__doc__
+    kneighbors.__doc__ = sklearn_LocalOutlierFactor.kneighbors.__doc__

sklearnex/neighbors/common.py

@@ -137,6 +137,9 @@ class KNeighborsDispatchingBase:
             self.n_features_in_ = X.data.shape[1]
 
     def _onedal_supported(self, device, method_name, *data):
+        if method_name == "fit":
+            self._fit_validation(data[0], data[1])
+
         class_name = self.__class__.__name__
         is_classifier = "Classifier" in class_name
         is_regressor = "Regressor" in class_name
@@ -249,7 +252,7 @@ class KNeighborsDispatchingBase:
                     class_count >= 2, "One-class case is not supported."
                 )
             return patching_status
-        if method_name in ["predict", "predict_proba", "kneighbors"]:
+        if method_name in ["predict", "predict_proba", "kneighbors", "score"]:
             patching_status.and_condition(
                 hasattr(self, "_onedal_estimator"), "oneDAL model was not trained."
             )