Fix sklearn tests [part 1] (#948)

Author: Pahandrovich

.circleci/deselect_tests.py

@@ -19,25 +19,25 @@
 import argparse
 import os.path
 from yaml import FullLoader, load as yaml_load
-from distutils.version import LooseVersion
+from packaging.version import Version
 import sklearn
 from sklearn import __version__ as sklearn_version
 import warnings
 
 
 def evaluate_cond(cond, v):
     if cond.startswith(">="):
-        return LooseVersion(v) >= LooseVersion(cond[2:])
+        return Version(v) >= Version(cond[2:])
     if cond.startswith("<="):
-        return LooseVersion(v) <= LooseVersion(cond[2:])
+        return Version(v) <= Version(cond[2:])
     if cond.startswith("=="):
-        return LooseVersion(v) == LooseVersion(cond[2:])
+        return Version(v) == Version(cond[2:])
     if cond.startswith("!="):
-        return LooseVersion(v) != LooseVersion(cond[2:])
+        return Version(v) != Version(cond[2:])
     if cond.startswith("<"):
-        return LooseVersion(v) < LooseVersion(cond[1:])
+        return Version(v) < Version(cond[1:])
     if cond.startswith(">"):
-        return LooseVersion(v) > LooseVersion(cond[1:])
+        return Version(v) > Version(cond[1:])
     warnings.warn(
         'Test selection condition "{0}" should start with '
         '>=, <=, ==, !=, < or > to compare to version of scikit-learn run. '

daal4py/sklearn/_utils.py

@@ -19,7 +19,7 @@
 
 from daal4py import _get__daal_link_version__ as dv
 from sklearn import __version__ as sklearn_version
-from distutils.version import LooseVersion
+from packaging.version import Version
 import logging
 
 
@@ -56,7 +56,7 @@ def daal_check_version(rule):
 
 
 def sklearn_check_version(ver):
-    return bool(LooseVersion(sklearn_version) >= LooseVersion(ver))
+    return bool(Version(sklearn_version) >= Version(ver))
 
 
 def get_daal_version():

daal4py/sklearn/cluster/_dbscan.py

@@ -24,8 +24,7 @@
 
 import daal4py
 from daal4py.sklearn._utils import (
-    make2d, getFPType, get_patch_message, PatchingConditionsChain)
-import logging
+    make2d, getFPType, PatchingConditionsChain)
 
 from .._device_offload import support_usm_ndarray
 from .._utils import sklearn_check_version

daal4py/sklearn/cluster/_k_means_0_22.py

@@ -35,9 +35,8 @@
 
 import daal4py
 from .._utils import (
-    getFPType, get_patch_message, daal_check_version, PatchingConditionsChain)
+    getFPType, daal_check_version, PatchingConditionsChain)
 from .._device_offload import support_usm_ndarray
-import logging
 
 
 def _tolerance(X, rtol):

daal4py/sklearn/cluster/_k_means_0_23.py

@@ -37,11 +37,9 @@
 import daal4py
 from .._utils import (
     getFPType,
-    get_patch_message,
     sklearn_check_version,
     PatchingConditionsChain)
 from .._device_offload import support_usm_ndarray
-import logging
 
 
 def _validate_center_shape(X, n_centers, centers):

daal4py/sklearn/decomposition/_pca.py

@@ -25,9 +25,8 @@
 
 import daal4py
 from .._utils import (
-    getFPType, get_patch_message, sklearn_check_version, PatchingConditionsChain)
+    getFPType, sklearn_check_version, PatchingConditionsChain)
 from .._device_offload import support_usm_ndarray
-import logging
 
 if sklearn_check_version('0.22'):
     from sklearn.decomposition._pca import PCA as PCA_original

daal4py/sklearn/ensemble/AdaBoostClassifier.py

@@ -26,7 +26,7 @@
 from .._utils import getFPType
 
 from sklearn import __version__ as sklearn_version
-from distutils.version import LooseVersion
+from packaging.version import Version
 
 
 class AdaBoostClassifier(BaseEstimator, ClassifierMixin):
@@ -129,7 +129,7 @@ def fit(self, X, y):
 
     def predict(self, X):
         # Check is fit had been called
-        if LooseVersion(sklearn_version) >= LooseVersion("0.22"):
+        if Version(sklearn_version) >= Version("0.22"):
             check_is_fitted(self)
         else:
             check_is_fitted(self, ['n_features_in_', 'n_classes_'])

daal4py/sklearn/ensemble/_forest.py

@@ -20,14 +20,14 @@
 import warnings
 
 import daal4py
-from .._utils import (getFPType, get_patch_message)
+from .._utils import getFPType
 from .._device_offload import support_usm_ndarray
 from daal4py.sklearn._utils import (
     daal_check_version, sklearn_check_version,
     PatchingConditionsChain)
 import logging
 
-from sklearn.tree import (DecisionTreeClassifier, DecisionTreeRegressor)
+from sklearn.tree import DecisionTreeClassifier
 from sklearn.tree._tree import Tree
 from sklearn.ensemble import RandomForestClassifier as RandomForestClassifier_original
 from sklearn.ensemble import RandomForestRegressor as RandomForestRegressor_original
@@ -40,8 +40,6 @@
 from sklearn.base import clone
 from sklearn.exceptions import DataConversionWarning
 
-from sklearn import __version__ as sklearn_version
-from distutils.version import LooseVersion
 from math import ceil
 from scipy import sparse as sp
 
@@ -97,6 +95,12 @@ def _get_n_samples_bootstrap(n_samples, max_samples):
 
 
 def _check_parameters(self):
+    if not self.bootstrap and self.max_samples is not None:
+        raise ValueError(
+            "`max_sample` cannot be set if `bootstrap=False`. "
+            "Either switch to `bootstrap=True` or set "
+            "`max_sample=None`."
+        )
     if isinstance(self.min_samples_leaf, numbers.Integral):
         if not 1 <= self.min_samples_leaf:
             raise ValueError("min_samples_leaf must be at least 1 "

daal4py/sklearn/linear_model/_linear.py

@@ -16,15 +16,11 @@
 
 import numpy as np
 from scipy import sparse as sp
-from scipy import linalg
 
-from sklearn.linear_model._base import _rescale_data
 from ..utils.validation import _daal_check_array, _daal_check_X_y
 from ..utils.base import _daal_validate_data
 from .._utils import sklearn_check_version
 from .._device_offload import support_usm_ndarray
-from sklearn.utils.fixes import sparse_lsqr
-from sklearn.utils.validation import _check_sample_weight
 from sklearn.utils import check_array
 
 from sklearn.linear_model import LinearRegression as LinearRegression_original

daal4py/sklearn/linear_model/logistic_path.py

@@ -48,9 +48,8 @@
     LogisticRegression as LogisticRegression_original)
 from sklearn.preprocessing import LabelEncoder, LabelBinarizer
 from .._utils import (
-    getFPType, get_patch_message, sklearn_check_version, PatchingConditionsChain)
+    getFPType, sklearn_check_version, PatchingConditionsChain)
 from .._device_offload import support_usm_ndarray
-import logging
 
 
 # Code adapted from sklearn.linear_model.logistic version 0.21