scikit-learn · rth · Jun 24, 2020 · Jun 23, 2020 · Jun 23, 2020 · Jun 23, 2020
diff --git a/conftest.py b/conftest.py
@@ -7,18 +7,18 @@
 
 import platform
 import sys
-from distutils.version import LooseVersion
 import os
 
 import pytest
 from _pytest.doctest import DoctestItem
 
 from sklearn.utils import _IS_32BIT
 from sklearn.externals import _pilutil
+from sklearn.utils.fixes import np_version, parse_version
 
 PYTEST_MIN_VERSION = '3.3.0'
 
-if LooseVersion(pytest.__version__) < PYTEST_MIN_VERSION:
+if parse_version(pytest.__version__) < parse_version(PYTEST_MIN_VERSION):
     raise ImportError('Your version of pytest is too old, you should have '
                       'at least pytest >= {} installed.'
                       .format(PYTEST_MIN_VERSION))
@@ -52,8 +52,7 @@ def pytest_collection_modifyitems(config, items):
     # run doctests only for numpy >= 1.14.
     skip_doctests = False
     try:
-        import numpy as np
-        if LooseVersion(np.__version__) < LooseVersion('1.14'):
+        if np_version < parse_version('1.14'):
             reason = 'doctests are only run for numpy >= 1.14'
             skip_doctests = True
         elif _IS_32BIT:

diff --git a/examples/cluster/plot_coin_segmentation.py b/examples/cluster/plot_coin_segmentation.py
@@ -25,7 +25,6 @@
 import time
 
 import numpy as np
-from distutils.version import LooseVersion
 from scipy.ndimage.filters import gaussian_filter
 import matplotlib.pyplot as plt
 import skimage
@@ -34,9 +33,10 @@
 
 from sklearn.feature_extraction import image
 from sklearn.cluster import spectral_clustering
+from sklearn.utils.fixes import parse_version
 
 # these were introduced in skimage-0.14
-if LooseVersion(skimage.__version__) >= '0.14':
+if parse_version(skimage.__version__) >= parse_version('0.14'):
     rescale_params = {'anti_aliasing': False, 'multichannel': False}
 else:
     rescale_params = {}

diff --git a/examples/cluster/plot_coin_ward_segmentation.py b/examples/cluster/plot_coin_ward_segmentation.py
@@ -17,7 +17,6 @@
 import time as time
 
 import numpy as np
-from distutils.version import LooseVersion
 from scipy.ndimage.filters import gaussian_filter
 
 import matplotlib.pyplot as plt
@@ -28,9 +27,10 @@
 
 from sklearn.feature_extraction.image import grid_to_graph
 from sklearn.cluster import AgglomerativeClustering
+from sklearn.utils.fixes import parse_version
 
 # these were introduced in skimage-0.14
-if LooseVersion(skimage.__version__) >= '0.14':
+if parse_version(skimage.__version__) >= parse_version('0.14'):
     rescale_params = {'anti_aliasing': False, 'multichannel': False}
 else:
     rescale_params = {}

diff --git a/examples/compose/plot_transformed_target.py b/examples/compose/plot_transformed_target.py
@@ -19,20 +19,20 @@
 import numpy as np
 import matplotlib
 import matplotlib.pyplot as plt
-from distutils.version import LooseVersion
 
 from sklearn.datasets import make_regression
 from sklearn.model_selection import train_test_split
 from sklearn.linear_model import RidgeCV
 from sklearn.compose import TransformedTargetRegressor
 from sklearn.metrics import median_absolute_error, r2_score
+from sklearn.utils.fixes import parse_version
 
 # %%
 # Synthetic example
 ##############################################################################
 
 # `normed` is being deprecated in favor of `density` in histograms
-if LooseVersion(matplotlib.__version__) >= '2.1':
+if parse_version(matplotlib.__version__) >= parse_version('2.1'):
     density_param = {'density': True}
 else:
     density_param = {'normed': True}

diff --git a/examples/decomposition/plot_sparse_coding.py b/examples/decomposition/plot_sparse_coding.py
@@ -16,12 +16,11 @@
 """
 print(__doc__)
 
-from distutils.version import LooseVersion
-
 import numpy as np
 import matplotlib.pyplot as plt
 
 from sklearn.decomposition import SparseCoder
+from sklearn.utils.fixes import np_version, parse_version
 
 
 def ricker_function(resolution, center, width):
@@ -68,7 +67,7 @@ def ricker_matrix(width, resolution, n_components):
               ('Lasso', 'lasso_lars', 2, None, 'turquoise'), ]
 lw = 2
 # Avoid FutureWarning about default value change when numpy >= 1.14
-lstsq_rcond = None if LooseVersion(np.__version__) >= '1.14' else -1
+lstsq_rcond = None if np_version >= parse_version('1.14') else -1
 
 plt.figure(figsize=(13, 6))
 for subplot, (D, title) in enumerate(zip((D_fixed, D_multi),

diff --git a/examples/miscellaneous/plot_johnson_lindenstrauss_bound.py b/examples/miscellaneous/plot_johnson_lindenstrauss_bound.py
@@ -19,15 +19,15 @@
 import numpy as np
 import matplotlib
 import matplotlib.pyplot as plt
-from distutils.version import LooseVersion
 from sklearn.random_projection import johnson_lindenstrauss_min_dim
 from sklearn.random_projection import SparseRandomProjection
 from sklearn.datasets import fetch_20newsgroups_vectorized
 from sklearn.datasets import load_digits
 from sklearn.metrics.pairwise import euclidean_distances
+from sklearn.utils.fixes import parse_version
 
 # `normed` is being deprecated in favor of `density` in histograms
-if LooseVersion(matplotlib.__version__) >= '2.1':
+if parse_version(matplotlib.__version__) >= parse_version('2.1'):
     density_param = {'density': True}
 else:
     density_param = {'normed': True}

diff --git a/examples/neighbors/plot_kde_1d.py b/examples/neighbors/plot_kde_1d.py
@@ -31,12 +31,12 @@
 import numpy as np
 import matplotlib
 import matplotlib.pyplot as plt
-from distutils.version import LooseVersion
 from scipy.stats import norm
 from sklearn.neighbors import KernelDensity
+from sklearn.utils.fixes import parse_version
 
 # `normed` is being deprecated in favor of `density` in histograms
-if LooseVersion(matplotlib.__version__) >= '2.1':
+if parse_version(matplotlib.__version__) >= parse_version('2.1'):
     density_param = {'density': True}
 else:
     density_param = {'normed': True}

diff --git a/sklearn/datasets/_lfw.py b/sklearn/datasets/_lfw.py
@@ -12,7 +12,6 @@
 from os.path import dirname, join, exists, isdir
 
 import logging
-from distutils.version import LooseVersion
 
 import numpy as np
 import joblib
@@ -21,6 +20,7 @@
 from ._base import get_data_home, _fetch_remote, RemoteFileMetadata
 from ..utils import Bunch
 from ..utils.validation import _deprecate_positional_args
+from ..utils.fixes import parse_version
 
 logger = logging.getLogger(__name__)
 
@@ -304,7 +304,7 @@ def fetch_lfw_people(*, data_home=None, funneled=True, resize=0.5,
 
     # wrap the loader in a memoizing function that will return memmaped data
     # arrays for optimal memory usage
-    if LooseVersion(joblib.__version__) < LooseVersion('0.12'):
+    if parse_version(joblib.__version__) < parse_version('0.12'):
         # Deal with change of API in joblib
         m = Memory(cachedir=lfw_home, compress=6, verbose=0)
     else:
@@ -477,7 +477,7 @@ def fetch_lfw_pairs(*, subset='train', data_home=None, funneled=True,
 
     # wrap the loader in a memoizing function that will return memmaped data
     # arrays for optimal memory usage
-    if LooseVersion(joblib.__version__) < LooseVersion('0.12'):
+    if parse_version(joblib.__version__) < parse_version('0.12'):
         # Deal with change of API in joblib
         m = Memory(cachedir=lfw_home, compress=6, verbose=0)
     else:

diff --git a/sklearn/ensemble/_hist_gradient_boosting/tests/test_loss.py b/sklearn/ensemble/_hist_gradient_boosting/tests/test_loss.py
@@ -4,7 +4,7 @@
 from scipy.optimize import newton
 from scipy.special import logit
 from sklearn.utils import assert_all_finite
-from sklearn.utils.fixes import sp_version
+from sklearn.utils.fixes import sp_version, parse_version
 import pytest
 
 from sklearn.ensemble._hist_gradient_boosting.loss import _LOSSES
@@ -60,7 +60,7 @@ def get_hessians(y_true, raw_predictions):
     ('poisson', 0., 2.),
     ('poisson', -22., 10.),
 ])
-@pytest.mark.skipif(sp_version == (1, 2, 0),
+@pytest.mark.skipif(sp_version == parse_version('1.2.0'),
                     reason='bug in scipy 1.2.0, see scipy issue #9608')
 @skip_if_32bit
 def test_derivatives(loss, x0, y_true):

diff --git a/sklearn/ensemble/tests/test_forest.py b/sklearn/ensemble/tests/test_forest.py
@@ -11,7 +11,6 @@
 import pickle
 import math
 from collections import defaultdict
-from distutils.version import LooseVersion
 import itertools
 from itertools import combinations
 from itertools import product
@@ -35,6 +34,7 @@
 from sklearn.utils._testing import assert_warns_message
 from sklearn.utils._testing import ignore_warnings
 from sklearn.utils._testing import skip_if_no_parallel
+from sklearn.utils.fixes import parse_version
 
 from sklearn.exceptions import NotFittedError
 
@@ -1274,7 +1274,7 @@ def start_call(self):
 joblib.register_parallel_backend('testing', MyBackend)
 
 
-@pytest.mark.skipif(joblib.__version__ < LooseVersion('0.12'),
+@pytest.mark.skipif(parse_version(joblib.__version__) < parse_version('0.12'),
                     reason='tests not yet supported in joblib <0.12')
 @skip_if_no_parallel
 def test_backend_respected():

diff --git a/sklearn/linear_model/tests/test_base.py b/sklearn/linear_model/tests/test_base.py
@@ -5,8 +5,6 @@
 
 import pytest
 
-from distutils.version import LooseVersion
-
 import numpy as np
 from scipy import sparse
 from scipy import linalg
@@ -15,6 +13,7 @@
 from sklearn.utils._testing import assert_array_equal
 from sklearn.utils._testing import assert_almost_equal
 from sklearn.utils._testing import assert_allclose
+from sklearn.utils.fixes import parse_version
 
 from sklearn.linear_model import LinearRegression
 from sklearn.linear_model._base import _preprocess_data
@@ -210,7 +209,7 @@ def test_linear_regression_sparse_multiple_outcome(random_state=0):
 def test_linear_regression_pd_sparse_dataframe_warning():
     pd = pytest.importorskip('pandas')
     # restrict the pd versions < '0.24.0' as they have a bug in is_sparse func
-    if LooseVersion(pd.__version__) < '0.24.0':
+    if parse_version(pd.__version__) < parse_version('0.24.0'):
         pytest.skip("pandas 0.24+ required.")
 
     # Warning is raised only when some of the columns is sparse

diff --git a/sklearn/linear_model/tests/test_coordinate_descent.py b/sklearn/linear_model/tests/test_coordinate_descent.py
@@ -8,7 +8,6 @@
 from scipy import interpolate, sparse
 from copy import deepcopy
 import joblib
-from distutils.version import LooseVersion
 
 from sklearn.datasets import load_diabetes
 from sklearn.datasets import make_regression
@@ -23,6 +22,7 @@
 from sklearn.utils._testing import ignore_warnings
 from sklearn.utils._testing import assert_array_equal
 from sklearn.utils._testing import TempMemmap
+from sklearn.utils.fixes import parse_version
 
 from sklearn.linear_model import Lasso, \
     LassoCV, ElasticNet, ElasticNetCV, MultiTaskLasso, MultiTaskElasticNet, \
@@ -1034,7 +1034,8 @@ def test_linear_models_cv_fit_for_all_backends(backend, estimator):
     # memmapped when using loky backend, causing an error due to unexpected
     # behavior of fancy indexing of read-only memmaps (cf. numpy#14132).
 
-    if joblib.__version__ < LooseVersion('0.12') and backend == 'loky':
+    if (parse_version(joblib.__version__) < parse_version('0.12')
+            and backend == 'loky'):
         pytest.skip('loky backend does not exist in joblib <0.12')
 
     # Create a problem sufficiently large to cause memmapping (1MB).

diff --git a/sklearn/linear_model/tests/test_least_angle.py b/sklearn/linear_model/tests/test_least_angle.py
@@ -1,7 +1,5 @@
 import warnings
 
-from distutils.version import LooseVersion
-
 import numpy as np
 import pytest
 from scipy import linalg
@@ -14,6 +12,7 @@
 from sklearn.utils._testing import ignore_warnings
 from sklearn.utils._testing import assert_warns
 from sklearn.utils._testing import TempMemmap
+from sklearn.utils.fixes import np_version, parse_version
 from sklearn.exceptions import ConvergenceWarning
 from sklearn import linear_model, datasets
 from sklearn.linear_model._least_angle import _lars_path_residues
@@ -123,7 +122,7 @@ def test_lars_lstsq():
     clf = linear_model.LassoLars(alpha=0.)
     clf.fit(X1, y)
     # Avoid FutureWarning about default value change when numpy >= 1.14
-    rcond = None if LooseVersion(np.__version__) >= '1.14' else -1
+    rcond = None if np_version >= parse_version('1.14') else -1
     coef_lstsq = np.linalg.lstsq(X1, y, rcond=rcond)[0]
     assert_array_almost_equal(clf.coef_, coef_lstsq)
 

diff --git a/sklearn/linear_model/tests/test_sgd.py b/sklearn/linear_model/tests/test_sgd.py
@@ -1,4 +1,3 @@
-from distutils.version import LooseVersion
 import pickle
 import pytest
 
@@ -13,6 +12,7 @@
 from sklearn.utils._testing import assert_raises_regexp
 from sklearn.utils._testing import assert_warns
 from sklearn.utils._testing import ignore_warnings
+from sklearn.utils.fixes import parse_version
 
 from sklearn import linear_model, datasets, metrics
 from sklearn.base import clone, is_classifier
@@ -1594,7 +1594,8 @@ def test_SGDClassifier_fit_for_all_backends(backend):
     # a segmentation fault when trying to write in a readonly memory mapped
     # buffer.
 
-    if joblib.__version__ < LooseVersion('0.12') and backend == 'loky':
+    if (parse_version(joblib.__version__) < parse_version('0.12')
+            and backend == 'loky'):
         pytest.skip('loky backend does not exist in joblib <0.12')
 
     random_state = np.random.RandomState(42)

diff --git a/sklearn/neighbors/_base.py b/sklearn/neighbors/_base.py
@@ -7,7 +7,6 @@
 #
 # License: BSD 3 clause (C) INRIA, University of Amsterdam
 from functools import partial
-from distutils.version import LooseVersion
 
 import warnings
 from abc import ABCMeta, abstractmethod
@@ -28,6 +27,7 @@
 from ..utils.multiclass import check_classification_targets
 from ..utils.validation import check_is_fitted
 from ..utils.validation import check_non_negative
+from ..utils.fixes import parse_version
 from ..exceptions import DataConversionWarning, EfficiencyWarning
 
 VALID_METRICS = dict(ball_tree=BallTree.valid_metrics,
@@ -650,7 +650,7 @@ class from an array representing our data set and ask who's
                     "%s does not work with sparse matrices. Densify the data, "
                     "or set algorithm='brute'" % self._fit_method)
             old_joblib = (
-                    LooseVersion(joblib.__version__) < LooseVersion('0.12'))
+                    parse_version(joblib.__version__) < parse_version('0.12'))
             if old_joblib:
                 # Deal with change of API in joblib
                 check_pickle = False if old_joblib else None
@@ -957,7 +957,7 @@ class from an array representing our data set and ask who's
                     "or set algorithm='brute'" % self._fit_method)
 
             n_jobs = effective_n_jobs(self.n_jobs)
-            if LooseVersion(joblib.__version__) < LooseVersion('0.12'):
+            if parse_version(joblib.__version__) < parse_version('0.12'):
                 # Deal with change of API in joblib
                 delayed_query = delayed(_tree_query_radius_parallel_helper,
                                         check_pickle=False)

diff --git a/sklearn/neighbors/tests/test_dist_metrics.py b/sklearn/neighbors/tests/test_dist_metrics.py
@@ -6,13 +6,12 @@
 
 import pytest
 
-from distutils.version import LooseVersion
-from scipy import __version__ as scipy_version
 from scipy.spatial.distance import cdist
 from sklearn.neighbors import DistanceMetric
 from sklearn.neighbors import BallTree
 from sklearn.utils import check_random_state
 from sklearn.utils._testing import assert_raises_regex
+from sklearn.utils.fixes import sp_version, parse_version
 
 
 def dist_func(x1, x2, p):
@@ -106,7 +105,7 @@ def check_pdist_bool(metric, D_true):
     # Based on https://github.com/scipy/scipy/pull/7373
     # When comparing two all-zero vectors, scipy>=1.2.0 jaccard metric
     # was changed to return 0, instead of nan.
-    if metric == 'jaccard' and LooseVersion(scipy_version) < '1.2.0':
+    if metric == 'jaccard' and sp_version < parse_version('1.2.0'):
         D_true[np.isnan(D_true)] = 0
     assert_array_almost_equal(D12, D_true)