Closed
Description
Describe the bug
The following tests fail with an old version of scipy
test_set_estimator_dropfrom ensemble/tests/test_voting.pytest_warning_boundsfrom gaussian_process/tests/test_gpc.pytest_warning_boundsgaussian_process/tests/test_gpr.pytest_logistic_regression_path_convergence_faillinear_model/tests/test_logistic.py
This is due to the fact that some scipy.optimize DeprecationWarning are thrown. Filtering them with
@pytest.mark.filterwarnings("ignore::DeprecationWarning") seems not to prevent record to remember those DeprecationWarning.
New scipy versions are not affected by this bug but the goal of this tests is to check the number of ConvergenceWarning on specific parameters and they should probably pass independently from other warnings.
Steps/Code to Reproduce
$ pytest sklearn/gaussian_process/tests/test_gpc.py
Expected Results
The test pass
Actual Results
============================================================= FAILURES ==============================================================
________________________________________________________ test_warning_bounds ____________________
5D9E
____________________________________
def test_warning_bounds():
kernel = RBF(length_scale_bounds=[1e-5, 1e-3])
gpc = GaussianProcessClassifier(kernel=kernel)
assert_warns_message(ConvergenceWarning, "The optimal value found for "
"dimension 0 of parameter "
"length_scale is close to "
"the specified upper bound "
"0.001. Increasing the bound "
"and calling fit again may "
"find a better value.",
gpc.fit, X, y)
kernel_sum = (WhiteKernel(noise_level_bounds=[1e-5, 1e-3]) +
RBF(length_scale_bounds=[1e3, 1e5]))
gpc_sum = GaussianProcessClassifier(kernel=kernel_sum)
with pytest.warns(None) as record:
gpc_sum.fit(X, y)
> assert len(record) == 2
E assert 7 == 2
E + where 7 = len(WarningsChecker(record=True))
sklearn/gaussian_process/tests/test_gpc.py:206: AssertionError
Versions
pytest 6.1.1
System:
python: 3.7.9 (default, Aug 31 2020, 12:42:55) [GCC 7.3.0]
executable: /home/cmarmo/.conda/envs/sklearn-travis/bin/python
machine: Linux-5.8.13-200.fc32.x86_64-x86_64-with-fedora-32-Thirty_Two
Python dependencies:
pip: 20.2.4
setuptools: 50.3.0.post20201006
sklearn: 0.24.dev0
numpy: 1.19.2
scipy: 1.3.0
Cython: None
pandas: None
matplotlib: None
joblib: 0.17.0
threadpoolctl: 2.1.0