fix(sklearn): address deprecated arguments to check_array in 1.6.0+

optbinning/_sklearn_compat.py

@@ -0,0 +1,39 @@
+"""Compatibility with multiple sklearn versions."""
+
+import functools
+
+from packaging.version import parse, Version
+import sklearn
+from sklearn.utils import check_array as _check_array
+
+
+@functools.lru_cache(maxsize=0)
+def _sklearn_version() -> Version:
+    """Return the version of sklearn as a version tuple.
+
+    This function is cached to avoid calling it multiple times.
+    """
+    return parse(sklearn.__version__)
+
+
+@functools.lru_cache(maxsize=0)
+def _new_check_array_api():
+    """Sklearn uses the new api after version 1.6.0+
+
+    https://github.com/scikit-learn/scikit-learn/issues/29262
+    """
+    return _sklearn_version() >= parse("1.6.0")
+
+def check_array(*args, **kwargs):
+    """Wrapper around check array to preserve backwards compatibility.
+
+    https://github.com/scikit-learn/scikit-learn/issues/29262
+    """
+    if _new_check_array_api() is False:
+        return _check_array(*args, **kwargs)
+
+    # Only replace if it's in the kwargs
+    finite = kwargs.pop("force_all_finite", None)
+    if finite is not None:
+        kwargs["ensure_all_finite"] = finite
+    return _check_array(*args, **kwargs) 

optbinning/binning/binning.py

@@ -10,12 +10,12 @@
 
 import numpy as np
 
-from sklearn.utils import check_array
 
 import json
 
 from ..information import solver_statistics
 from ..logging import Logger
+from .._sklearn_compat import check_array
 from .auto_monotonic import auto_monotonic
 from .auto_monotonic import peak_valley_trend_change_heuristic
 from .base import BaseOptimalBinning

optbinning/binning/binning_process.py

@@ -17,11 +17,11 @@
 from joblib import Parallel, delayed, effective_n_jobs
 from sklearn.base import BaseEstimator
 from sklearn.exceptions import NotFittedError
-from sklearn.utils import check_array
 from sklearn.utils import check_consistent_length
 from sklearn.utils.multiclass import type_of_target
 
 from ..logging import Logger
+from .._sklearn_compat import check_array
 from .base import Base
 from .binning import OptimalBinning
 from .binning_process_information import print_binning_process_information

optbinning/binning/continuous_binning.py

@@ -9,12 +9,12 @@
 import time
 import json
 
-from sklearn.utils import check_array
 
 import numpy as np
 
 from ..information import solver_statistics
 from ..logging import Logger
+from .._sklearn_compat import check_array
 from .auto_monotonic import auto_monotonic_continuous
 from .auto_monotonic import peak_valley_trend_change_heuristic
 from .binning import OptimalBinning

optbinning/binning/mdlp.py

@@ -12,7 +12,8 @@
 from scipy import special
 from sklearn.base import BaseEstimator
 from sklearn.exceptions import NotFittedError
-from sklearn.utils import check_array
+
+from .._sklearn_compat import check_array
 
 
 def _check_parameters(min_samples_split, min_samples_leaf, max_candidates):

optbinning/binning/metrics.py

@@ -9,9 +9,10 @@
 
 from scipy import special
 from scipy import stats
-from sklearn.utils import check_array
 from sklearn.utils import check_consistent_length
 
+from .._sklearn_compat import check_array
+
 
 def _check_x_y(x, y):
     x = check_array(x, ensure_2d=False, force_all_finite=True)

optbinning/binning/multiclass_binning.py

@@ -11,10 +11,10 @@
 
 import numpy as np
 
-from sklearn.utils import check_array
 
 from ..information import solver_statistics
 from ..logging import Logger
+from .._sklearn_compat import check_array
 from .auto_monotonic import auto_monotonic
 from .auto_monotonic import peak_valley_trend_change_heuristic
 from .binning import OptimalBinning

optbinning/binning/multidimensional/preprocessing_2d.py

@@ -8,9 +8,9 @@
 import numpy as np
 import pandas as pd
 
-from sklearn.utils import check_array
 from sklearn.utils import check_consistent_length
 
+from ..._sklearn_compat import check_array
 from ..preprocessing import categorical_transform
 
 

optbinning/binning/multidimensional/transformations_2d.py

@@ -8,8 +8,9 @@
 import numpy as np
 import pandas as pd
 
-from sklearn.utils import check_array
 
+
+from ..._sklearn_compat import check_array
 from ..transformations import _check_metric_special_missing
 from ..transformations import _check_show_digits
 from .binning_statistics_2d import bin_categorical

optbinning/binning/piecewise/transformations.py

@@ -8,11 +8,11 @@
 import numpy as np
 import pandas as pd
 
-from sklearn.utils import check_array
 
 from ...binning.transformations import transform_event_rate_to_woe
 from ...binning.transformations import _check_metric_special_missing
 from ...binning.transformations import _mask_special_missing
+from ..._sklearn_compat import check_array
 
 
 def _apply_transform(x, c, lb, ub, special_codes, metric_special,

optbinning/binning/preprocessing.py

@@ -11,11 +11,11 @@
 import pandas as pd
 
 from sklearn.preprocessing import LabelEncoder
-from sklearn.utils import check_array
 from sklearn.utils import check_consistent_length
 from sklearn.utils import compute_class_weight
 from sklearn.utils.validation import _check_sample_weight
 
+from .._sklearn_compat import check_array
 from .outlier import ModifiedZScoreDetector
 from .outlier import RangeDetector
 from .outlier import YQuantileDetector

optbinning/binning/transformations.py

@@ -10,8 +10,7 @@
 import numpy as np
 import pandas as pd
 
-from sklearn.utils import check_array
-
+from .._sklearn_compat import check_array
 from .binning_statistics import bin_categorical
 from .binning_statistics import bin_str_format
 

optbinning/binning/uncertainty/binning_scenarios.py

@@ -11,11 +11,11 @@
 
 import numpy as np
 
-from sklearn.utils import check_array
 
 from ...information import solver_statistics
 from ...logging import Logger
 from ...binning.preprocessing import split_data_scenarios
+from ..._sklearn_compat import check_array
 from ..binning import OptimalBinning
 from ..binning_statistics import bin_info
 from ..binning_statistics import BinningTable

optbinning/scorecard/plots.py

@@ -11,9 +11,10 @@
 import matplotlib.pyplot as plt
 
 from sklearn.metrics import roc_curve, roc_auc_score
-from sklearn.utils import check_array
 from sklearn.utils import check_consistent_length
 
+from .._sklearn_compat import check_array
+
 
 def _check_arrays(y, y_pred):
     y = check_array(y, ensure_2d=False, force_all_finite=True)

setup.py

@@ -36,6 +36,7 @@ def run(self):
     'matplotlib',
     'numpy>=1.16.1',
     'ortools>=9.4,<9.12',
+    'packaging',
     'pandas',
     'ropwr>=1.0.0',
     'scikit-learn>=1.0.2',

tests/test_sklearn_compat.py

@@ -0,0 +1,51 @@
+"""Tests for sklearn compat."""
+
+from unittest import mock
+
+import pytest
+
+from optbinning import _sklearn_compat
+
+
+@pytest.fixture()
+def _cache_clear_version():
+    """Clear the cache each time it's called so multiple tests can run."""
+    _sklearn_compat._sklearn_version.cache_clear()
+    yield
+    _sklearn_compat._sklearn_version.cache_clear()
+
+
+@pytest.fixture()
+def _cache_clear_new_check_array_api():
+    """Clear the cache each time it's called so multiple tests can run."""
+    _sklearn_compat._new_check_array_api.cache_clear()
+    yield
+    _sklearn_compat._new_check_array_api.cache_clear()
+
+
+@pytest.mark.parametrize(
+    ("sklearn_version", "want"),
+    [
+        ("1.0.2", {"force_all_finite": True}),
+        # post releases
+        ("1.0.2.post", {"force_all_finite": True}),
+        # dev releases
+        ("1.0.dev0", {"force_all_finite": True}),
+        ("1.6.0", {"ensure_all_finite": True}),
+        ("1.7.0", {"ensure_all_finite": True}),
+        # release candidate
+        ("1.7.0rc1", {"ensure_all_finite": True}),
+    ],
+)
+def test__check_array_ensure_finite_kwargs(
+    sklearn_version, want, _cache_clear_version, _cache_clear_new_check_array_api
+):
+    with mock.patch.object(
+        _sklearn_compat.sklearn, "__version__", new=sklearn_version
+    ), mock.patch.object(
+        _sklearn_compat, "_check_array", return_value=""
+    ) as mock_check_array:
+        _sklearn_compat.check_array(force_all_finite=True)
+        kwargs = mock_check_array.call_args.kwargs
+        assert kwargs == want
+