MAINT run latest CI on Python 3.8 (scikit-learn#15637)

Author: amueller

azure-pipelines.yml

@@ -68,7 +68,7 @@ jobs:
       pylatest_pip_openblas_pandas:
         DISTRIB: 'conda-pip-latest'
         # FIXME: pinned until SciPy wheels are available for Pyhon 3.8
-        PYTHON_VERSION: '3.7'
+        PYTHON_VERSION: '3.8'
         PYTEST_VERSION: '4.6.2'
         COVERAGE: 'true'
         CHECK_PYTEST_SOFT_DEPENDENCY: 'true'

sklearn/ensemble/tests/test_partial_dependence.py

@@ -15,6 +15,12 @@
 from sklearn.utils._testing import ignore_warnings
 
 
+# TODO: Remove when https://github.com/numpy/numpy/issues/14397 is resolved
+pytestmark = pytest.mark.filterwarnings(
+    "ignore:In future, it will be an error for 'np.bool_':DeprecationWarning:"
+    "matplotlib.*")
+
+
 # toy sample
 X = [[-2, -1], [-1, -1], [-1, -2], [1, 1], [1, 2], [2, 1]]
 y = [-1, -1, -1, 1, 1, 1]

sklearn/inspection/tests/test_plot_partial_dependence.py

@@ -15,6 +15,12 @@
 from sklearn.inspection import plot_partial_dependence
 
 
+# TODO: Remove when https://github.com/numpy/numpy/issues/14397 is resolved
+pytestmark = pytest.mark.filterwarnings(
+    "ignore:In future, it will be an error for 'np.bool_':DeprecationWarning:"
+    "matplotlib.*")
+
+
 @pytest.fixture(scope="module")
 def boston():
     return load_boston()

sklearn/linear_model/tests/test_bayes.py

@@ -209,7 +209,7 @@ def test_ard_accuracy_on_easy_problem():
     X = np.random.RandomState(seed=seed).normal(size=(250, 3))
     y = X[:, 1]
 
-    regressor = ARDRegression()
+    regressor = ARDRegression(n_iter=600)
     regressor.fit(X, y)
 
     abs_coef_error = np.abs(1 - regressor.coef_[1])

sklearn/metrics/_plot/tests/test_plot_confusion_matrix.py

@@ -16,6 +16,11 @@
 from sklearn.metrics import ConfusionMatrixDisplay
 
 
+# TODO: Remove when https://github.com/numpy/numpy/issues/14397 is resolved
+pytestmark = pytest.mark.filterwarnings(
+    "ignore:In future, it will be an error for 'np.bool_':DeprecationWarning:"
+    "matplotlib.*")
+
 @pytest.fixture(scope="module")
 def n_classes():
     return 5

sklearn/metrics/_plot/tests/test_plot_precision_recall.py

@@ -15,6 +15,12 @@
 from sklearn.compose import make_column_transformer
 
 
+# TODO: Remove when https://github.com/numpy/numpy/issues/14397 is resolved
+pytestmark = pytest.mark.filterwarnings(
+    "ignore:In future, it will be an error for 'np.bool_':DeprecationWarning:"
+    "matplotlib.*")
+
+
 def test_errors(pyplot):
     X, y_multiclass = make_classification(n_classes=3, n_samples=50,
                                           n_informative=3,

sklearn/metrics/_plot/tests/test_plot_roc_curve.py

@@ -14,6 +14,12 @@
 from sklearn.compose import make_column_transformer
 
 
+# TODO: Remove when https://github.com/numpy/numpy/issues/14397 is resolved
+pytestmark = pytest.mark.filterwarnings(
+    "ignore:In future, it will be an error for 'np.bool_':DeprecationWarning:"
+    "matplotlib.*")
+
+
 @pytest.fixture(scope="module")
 def data():
     return load_iris(return_X_y=True)

sklearn/neighbors/_base.py

@@ -262,7 +262,7 @@ def _radius_neighbors_from_graph(graph, radius, return_distance):
     """
     assert graph.format == 'csr'
 
-    no_filter_needed = graph.data.max() <= radius
+    no_filter_needed = bool(graph.data.max() <= radius)
 
     if no_filter_needed:
         data, indices, indptr = graph.data, graph.indices, graph.indptr