sklearn tests subset in CI (#1199)

* Subset of sklearn tests in CI

* New deselected tests

* Filtering of deselected tests based on sys.platform

* Pinning of scipy version in CI to prevent sklearn test failures related to scipy API and impl. changes

* Parameters validation from sklearn 1.1 integrated to sklearnex DBSCAN, SVM and linear models.

Author: Alexsandruss

.ci/pipeline/ci.yml

@@ -101,7 +101,7 @@ jobs:
   - script: |
       conda update -y -q conda
       if [ $(echo $(PYTHON_VERSION) | grep '3.11') ]; then export DPCTL_PACKAGE=; else export DPCTL_PACKAGE=; fi
-      conda create -q -y -n CB -c conda-forge -c intel python=$(PYTHON_VERSION) dal-devel mpich $DPCTL_PACKAGE
+      conda create -q -y -n CB -c conda-forge -c intel python=$(PYTHON_VERSION) dal-devel mpich pyyaml $DPCTL_PACKAGE
     displayName: 'Conda create'
   - script: |
       . /usr/share/miniconda/etc/profile.d/conda.sh
@@ -128,16 +128,21 @@ jobs:
       conda activate CB
       sed -i -e "s/scikit-learn==1.2.0/scikit-learn==$(SKLEARN_VERSION).*/" requirements-test.txt
       sed -i -e "s/scikit-learn==1.0.2/scikit-learn==$(SKLEARN_VERSION).*/" requirements-test.txt
-      pip install --upgrade -r requirements-test.txt
-      pip install -r requirements-test-optional.txt
+      pip install --upgrade -r requirements-test.txt -r requirements-test-optional.txt
+      pip install $(python .ci/scripts/get_compatible_scipy_version.py)
       pip list
     displayName: 'Install testing requirements'
   - script: |
       . /usr/share/miniconda/etc/profile.d/conda.sh
       conda activate CB
       cd ..
       ./s/conda-recipe/run_test.sh
-    displayName: 'Testing'
+    displayName: 'Sklearnex testing'
+  - script: |
+      . /usr/share/miniconda/etc/profile.d/conda.sh
+      conda activate CB
+      bash .ci/scripts/run_sklearn_tests.sh
+    displayName: 'Sklearn testing'
   - script: |
       . /usr/share/miniconda/etc/profile.d/conda.sh
       conda activate CB
@@ -170,7 +175,7 @@ jobs:
       sudo chown -R $USER $CONDA
       conda config --set always_yes yes --set changeps1 no
       conda update -q conda
-      conda create -n CB -c conda-forge python=$(PYTHON_VERSION) dal dal-include mpich clang-format
+      conda create -n CB -c conda-forge python=$(PYTHON_VERSION) dal dal-include mpich clang-format pyyaml
     displayName: Create Anaconda environment
   - script: |
       source activate CB
@@ -192,15 +197,19 @@ jobs:
       source activate CB
       sed -i.bak -e "s/scikit-learn==1.2.0/scikit-learn==$(SKLEARN_VERSION).*/" requirements-test.txt
       sed -i.bak -e "s/scikit-learn==1.0.2/scikit-learn==$(SKLEARN_VERSION).*/" requirements-test.txt
-      pip install --upgrade -r requirements-test.txt
-      pip install -r requirements-test-optional.txt
+      pip install --upgrade -r requirements-test.txt -r requirements-test-optional.txt
+      pip install $(python .ci/scripts/get_compatible_scipy_version.py)
       pip list
     displayName: 'Install testing requirements'
   - script: |
       source activate CB
       cd ..
       ./s/conda-recipe/run_test.sh
-    displayName: Testing
+    displayName: 'Sklearnex testing'
+  - script: |
+      source activate CB
+      bash .ci/scripts/run_sklearn_tests.sh
+    displayName: 'Sklearn testing'
   - script: |
       source activate CB
       cd ..
@@ -229,7 +238,7 @@ jobs:
   steps:
   - powershell: Write-Host "##vso[task.prependpath]$env:CONDA\Scripts"
     displayName: Add conda to PATH
-  - script: conda create -q -y -n CB -c conda-forge -c intel python=$(PYTHON_VERSION) dal-devel impi-devel clang-format
+  - script: conda create -q -y -n CB -c conda-forge -c intel python=$(PYTHON_VERSION) dal-devel impi-devel clang-format pyyaml
     displayName: 'Create Anaconda environment'
   - script: |
       call activate CB
@@ -257,15 +266,21 @@ jobs:
       call activate CB
       sed -i -e "s/scikit-learn==1.2.0/scikit-learn==$(SKLEARN_VERSION).*/" requirements-test.txt
       sed -i -e "s/scikit-learn==1.0.2/scikit-learn==$(SKLEARN_VERSION).*/" requirements-test.txt
-      pip install --upgrade -r requirements-test.txt
-      pip install -r requirements-test-optional.txt
+      pip install --upgrade -r requirements-test.txt -r requirements-test-optional.txt
+      cd ..
+      for /f "delims=" %%c in ('python s\.ci\scripts\get_compatible_scipy_version.py') do set SCIPY_VERSION=%%c
+      pip install %SCIPY_VERSION%
       pip list
     displayName: 'Install testing requirements'
   - script: |
       call activate CB
       cd ..
       call s\conda-recipe\run_test.bat s
-    displayName: daal4py/sklearnex testing
+    displayName: 'Sklearnex testing'
+  - script: |
+      call activate CB
+      bash .ci/scripts/run_sklearn_tests.sh
+    displayName: 'Sklearn testing'
   - script: |
       call activate CB
       cd ..

.ci/scripts/get_compatible_scipy_version.py

@@ -0,0 +1,29 @@
+#! /usr/bin/env python
+#===============================================================================
+# Copyright 2023 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.
+#===============================================================================
+from daal4py.sklearn._utils import sklearn_check_version
+
+
+if sklearn_check_version('1.2'):
+    print('scipy==1.9')
+elif sklearn_check_version('1.1'):
+    print('scipy==1.8')
+elif sklearn_check_version('1.0'):
+    print('scipy==1.7')
+elif sklearn_check_version('0.24'):
+    print('scipy==1.6')
+else:
+    print('scipy')

.ci/scripts/run_sklearn_tests.sh

@@ -0,0 +1,25 @@
+#!/bin/bash
+#===============================================================================
+# Copyright 2023 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.
+#===============================================================================
+
+ci_dir=$( dirname $( dirname "${BASH_SOURCE[0]}" ) )
+cd $ci_dir
+
+export SELECTED_TESTS=$(python scripts/select_sklearn_tests.py)
+export DESELECTED_TESTS=$(python ../.circleci/deselect_tests.py ../deselected_tests.yaml)
+cd $(python -c "import sklearn, os; print(os.path.dirname(sklearn.__file__))")
+export SKLEARNEX_VERBOSE=DEBUG
+python -m sklearnex -m pytest --verbose --pyargs --durations=100 --durations-min=0.01 $DESELECTED_TESTS $SELECTED_TESTS

.ci/scripts/select_sklearn_tests.py

@@ -0,0 +1,79 @@
+#! /usr/bin/env python
+#===============================================================================
+# Copyright 2023 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 argparse
+import sys
+
+
+def parse_tests_tree(entry, prefix=''):
+    global tests_list
+
+    if isinstance(entry, dict):
+        for key, value in entry.items():
+            parse_tests_tree(value, f'{prefix}/{key}' if prefix != '' else key)
+    elif isinstance(entry, list):
+        for value in entry:
+            parse_tests_tree(value, prefix)
+    elif isinstance(entry, str):
+        tests_list.append(f'{prefix}/{entry}' if prefix != '' else entry)
+    else:
+        raise ValueError(f'Unknown type {type(entry)} in tests map')
+
+
+tests_map = {
+    'cluster/tests': [
+        'test_dbscan.py',
+        'test_k_means.py'
+    ],
+    'decomposition/tests': 'test_pca.py',
+    'ensemble/tests': 'test_forest.py',
+    'linear_model/tests': [
+        'test_base.py',
+        'test_coordinate_descent.py',
+        'test_ridge.py'
+    ],
+    'manifold/tests': 'test_t_sne.py',
+    'model_selection/tests': [
+        'test_split.py',
+        'test_validation.py'
+    ],
+    'neighbors/tests': [
+        'test_lof.py',
+        'test_neighbors.py',
+        'test_neighbors_pipeline.py'
+    ],
+    'svm/tests': [
+        'test_sparse.py',
+        'test_svm.py'
+    ]
+}
+
+
+if __name__ == '__main__':
+    arg_parser = argparse.ArgumentParser()
+    arg_parser.add_argument('--base-dir', type=str, default='')
+    args = arg_parser.parse_args()
+
+    tests_list = []
+    parse_tests_tree(tests_map, args.base_dir)
+    result = ''
+    for test in tests_list:
+
+        result += test + ' '
+    # correct paths for non-Unix envs
+    if sys.platform in ['win32', 'cygwin']:
+        result = result.replace('/', '\\')
+    print(result[:-1])

.circleci/deselect_tests.py

@@ -17,6 +17,7 @@
 
 # coding: utf-8
 import argparse
+import sys
 import os.path
 from yaml import FullLoader, load as yaml_load
 try:
@@ -48,15 +49,19 @@ def evaluate_cond(cond, v):
     return False
 
 
-def filter_by_version(entry, sk_ver):
+def filter_by_version_and_platform(entry, sk_ver):
     if not entry:
         return None
     t = entry.split(' ')
     if len(t) == 1:
         return entry
-    if len(t) != 2:
+    elif len(t) == 2:
+        t.append(None)
+    if len(t) != 3:
+        return None
+    test_name, cond, platform = t
+    if platform is not None and platform != sys.platform:
         return None
-    test_name, cond = t
     conds = cond.split(',')
     if all([evaluate_cond(cond, sk_ver) for cond in conds]):
         return test_name
@@ -79,19 +84,19 @@ def create_pytest_switches(filename, absolute, reduced, public, gpu, base_dir=No
             base_dir += '/'
 
         filtered_deselection = [
-            filter_by_version(test_name, sklearn_version)
+            filter_by_version_and_platform(test_name, sklearn_version)
             for test_name in dt.get('deselected_tests', [])]
         if reduced:
             filtered_deselection.extend(
-                [filter_by_version(test_name, sklearn_version)
+                [filter_by_version_and_platform(test_name, sklearn_version)
                  for test_name in dt.get('reduced_tests', [])])
         if public:
             filtered_deselection.extend(
-                [filter_by_version(test_name, sklearn_version)
+                [filter_by_version_and_platform(test_name, sklearn_version)
                  for test_name in dt.get('public', [])])
         if gpu:
             filtered_deselection.extend(
-                [filter_by_version(test_name, sklearn_version)
+                [filter_by_version_and_platform(test_name, sklearn_version)
                  for test_name in dt.get('gpu', [])])
         pytest_switches = []
         for test_name in filtered_deselection:

daal4py/sklearn/cluster/_dbscan.py

@@ -16,6 +16,7 @@
 
 import numpy as np
 from scipy import sparse as sp
+import numbers
 
 from sklearn.utils import check_array
 from sklearn.utils.validation import _check_sample_weight
@@ -29,6 +30,9 @@
 from .._device_offload import support_usm_ndarray
 from .._utils import sklearn_check_version
 
+if sklearn_check_version('1.1') and not sklearn_check_version('1.2'):
+    from sklearn.utils import check_scalar
+
 
 def _daal_dbscan(X, eps=0.5, min_samples=5, sample_weight=None):
     ww = make2d(sample_weight) if sample_weight is not None else None
@@ -236,6 +240,38 @@ def fit(self, X, y=None, sample_weight=None):
         """
         if sklearn_check_version("1.2"):
             self._validate_params()
+        elif sklearn_check_version("1.1"):
+            check_scalar(
+                self.eps,
+                "eps",
+                target_type=numbers.Real,
+                min_val=0.0,
+                include_boundaries="neither",
+            )
+            check_scalar(
+                self.min_samples,
+                "min_samples",
+                target_type=numbers.Integral,
+                min_val=1,
+                include_boundaries="left",
+            )
+            check_scalar(
+                self.leaf_size,
+                "leaf_size",
+                target_type=numbers.Integral,
+                min_val=1,
+                include_boundaries="left",
+            )
+            if self.p is not None:
+                check_scalar(
+                    self.p,
+                    "p",
+                    target_type=numbers.Real,
+                    min_val=0.0,
+                    include_boundaries="left",
+                )
+            if self.n_jobs is not None:
+                check_scalar(self.n_jobs, "n_jobs", target_type=numbers.Integral)
         else:
             if self.eps <= 0.0:
                 raise ValueError(f"eps == {self.eps}, must be > 0.0.")

daal4py/sklearn/linear_model/_coordinate_descent.py

@@ -25,6 +25,8 @@
     make2d, getFPType, get_patch_message, sklearn_check_version, PatchingConditionsChain)
 if sklearn_check_version('1.0') and not sklearn_check_version('1.2'):
     from sklearn.linear_model._base import _deprecate_normalize
+if sklearn_check_version('1.1') and not sklearn_check_version('1.2'):
+    from sklearn.utils import check_scalar
 
 import logging
 
@@ -394,6 +396,37 @@ def _fit(self, X, y, sample_weight=None, check_input=True):
         self._check_feature_names(X, reset=True)
     if sklearn_check_version("1.2"):
         self._validate_params()
+    elif sklearn_check_version('1.1'):
+        check_scalar(
+            self.alpha,
+            "alpha",
+            target_type=numbers.Real,
+            min_val=0.0,
+        )
+        if self.alpha == 0:
+            warnings.warn(
+                "With alpha=0, this algorithm does not converge "
+                "well. You are advised to use the LinearRegression "
+                "estimator",
+                stacklevel=2,
+            )
+        if isinstance(self.precompute, str):
+            raise ValueError(
+                "precompute should be one of True, False or array-like. Got %r"
+                % self.precompute
+            )
+        check_scalar(
+            self.l1_ratio,
+            "l1_ratio",
+            target_type=numbers.Real,
+            min_val=0.0,
+            max_val=1.0,
+        )
+        if self.max_iter is not None:
+            check_scalar(
+                self.max_iter, "max_iter", target_type=numbers.Integral, min_val=1
+            )
+        check_scalar(self.tol, "tol", target_type=numbers.Real, min_val=0.0)
     # check X and y
     if check_input:
         X, y = check_X_y(