[enhancement] Enable oneDAL support for EmpiricalCovariance with default parameters (#1677)

* Update covariance.py

* Update covariance.py

* formatting

* Update covariance.py

* Update covariance.py

* Update covariance.py

* isort

* Update covariance.py

* Update covariance.py

* Update covariance.py

Author: icfaust

onedal/covariance/covariance.py

@@ -16,10 +16,10 @@
 from abc import ABCMeta
 
 import numpy as np
-from sklearn.utils import check_array
 
-from daal4py.sklearn._utils import daal_check_version, get_dtype, make2d
+from daal4py.sklearn._utils import daal_check_version, get_dtype
 from onedal import _backend
+from onedal.utils import _check_array
 
 from ..common._policy import _get_policy
 from ..common.hyperparameters import get_hyperparameters
@@ -86,11 +86,7 @@ def fit(self, X, queue=None):
             Returns the instance itself.
         """
         policy = self._get_policy(queue, X)
-        X = check_array(X, dtype=[np.float64, np.float32])
-        X = make2d(X)
-        types = [np.float32, np.float64]
-        if get_dtype(X) not in types:
-            X = X.astype(np.float64)
+        X = _check_array(X, dtype=[np.float64, np.float32])
         X = _convert_to_supported(policy, X)
         dtype = get_dtype(X)
         params = self._get_onedal_params(dtype)

sklearnex/preview/covariance/covariance.py

@@ -14,6 +14,9 @@
 # limitations under the License.
 # ===============================================================================
 
+import warnings
+
+import numpy as np
 from scipy import sparse as sp
 from sklearn.covariance import EmpiricalCovariance as sklearn_EmpiricalCovariance
 from sklearn.utils import check_array
@@ -22,13 +25,15 @@
 from daal4py.sklearn._utils import sklearn_check_version
 from onedal.common.hyperparameters import get_hyperparameters
 from onedal.covariance import EmpiricalCovariance as onedal_EmpiricalCovariance
+from sklearnex import config_context
+from sklearnex.metrics import pairwise_distances
 
-from ..._device_offload import dispatch
+from ..._device_offload import dispatch, wrap_output_data
 from ..._utils import PatchingConditionsChain, register_hyperparameters
 
 
 @register_hyperparameters({"fit": get_hyperparameters("covariance", "compute")})
-@control_n_jobs(decorated_methods=["fit"])
+@control_n_jobs(decorated_methods=["fit", "mahalanobis"])
 class EmpiricalCovariance(sklearn_EmpiricalCovariance):
     __doc__ = sklearn_EmpiricalCovariance.__doc__
 
@@ -39,12 +44,17 @@ class EmpiricalCovariance(sklearn_EmpiricalCovariance):
 
     def _save_attributes(self):
         assert hasattr(self, "_onedal_estimator")
-        self.covariance_ = self._onedal_estimator.covariance_
+        self._set_covariance(self._onedal_estimator.covariance_)
         self.location_ = self._onedal_estimator.location_
 
     _onedal_covariance = staticmethod(onedal_EmpiricalCovariance)
 
     def _onedal_fit(self, X, queue=None):
+        if X.shape[0] == 1:
+            warnings.warn(
+                "Only one sample available. You may want to reshape your data array"
+            )
+
         onedal_params = {
             "method": "dense",
             "bias": True,
@@ -59,18 +69,14 @@ def _onedal_supported(self, method_name, *data):
         patching_status = PatchingConditionsChain(
             f"sklearn.covariance.{class_name}.{method_name}"
         )
-        if method_name == "fit":
+        if method_name in ["fit", "mahalanobis"]:
             (X,) = data
             patching_status.and_conditions(
                 [
                     (
                         self.assume_centered == False,
                         "assume_centered parameter is not supported on oneDAL side",
                     ),
-                    (
-                        self.store_precision == False,
-                        "precision matrix calculation is not supported on oneDAL side",
-                    ),
                     (not sp.issparse(X), "X is sparse. Sparse input is not supported."),
                 ]
             )
@@ -84,9 +90,9 @@ def fit(self, X, y=None):
         if sklearn_check_version("1.2"):
             self._validate_params()
         if sklearn_check_version("0.23"):
-            self._validate_data(X)
+            X = self._validate_data(X, force_all_finite=False)
         else:
-            check_array(X)
+            X = check_array(X, force_all_finite=False)
 
         dispatch(
             self,
@@ -100,4 +106,27 @@ def fit(self, X, y=None):
 
         return self
 
+    # expose sklearnex pairwise_distances if mahalanobis distance eventually supported
+    @wrap_output_data
+    def mahalanobis(self, X):
+        if sklearn_check_version("1.0"):
+            X = self._validate_data(X, reset=False)
+        else:
+            X = check_array(X)
+
+        precision = self.get_precision()
+        with config_context(assume_finite=True):
+            # compute mahalanobis distances
+            dist = pairwise_distances(
+                X, self.location_[np.newaxis, :], metric="mahalanobis", VI=precision
+            )
+
+        return np.reshape(dist, (len(X),)) ** 2
+
+    error_norm = wrap_output_data(sklearn_EmpiricalCovariance.error_norm)
+    score = wrap_output_data(sklearn_EmpiricalCovariance.score)
+
     fit.__doc__ = sklearn_EmpiricalCovariance.fit.__doc__
+    mahalanobis.__doc__ = sklearn_EmpiricalCovariance.mahalanobis
+    error_norm.__doc__ = sklearn_EmpiricalCovariance.error_norm.__doc__
+    score.__doc__ = sklearn_EmpiricalCovariance.score.__doc__