Skip to content

TST use global_random_seed in sklearn/_loss/tests/test_link.py #23751

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 6 commits into from
Jun 30, 2022
Merged

TST use global_random_seed in sklearn/_loss/tests/test_link.py #23751

Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
d19e56b
test_link [all random seeds]
Jun 24, 2022
a9b543c
use global_random_seed % 10
Jun 28, 2022
58e3f32
revert changes in test_link_out_argument function
Jun 28, 2022
e10032f
draw samples from uniform distribution
Jun 29, 2022
5226eba
add documentation
Jun 30, 2022
22f9b2f
correct documentation
Jun 30, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
13 changes: 7 additions & 6 deletions sklearn/_loss/tests/test_link.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -58,20 +58,21 @@ def test_is_in_range(interval):


@pytest.mark.parametrize("link", LINK_FUNCTIONS)
def test_link_inverse_identity(link):
def test_link_inverse_identity(link, global_random_seed):
# Test that link of inverse gives identity.
rng = np.random.RandomState(42)
rng = np.random.RandomState(global_random_seed)
link = link()
n_samples, n_classes = 100, None
# The values for `raw_prediction` are limited from -20 to 20 because in the
# class `LogitLink` the term `expit(x)` comes very close to 1 for large
# positive x and therefore loses precision.
if link.is_multiclass:
n_classes = 10
raw_prediction = rng.normal(loc=0, scale=10, size=(n_samples, n_classes))
raw_prediction = rng.uniform(low=-20, high=20, size=(n_samples, n_classes))
if isinstance(link, MultinomialLogit):
raw_prediction = link.symmetrize_raw_prediction(raw_prediction)
else:
# So far, the valid interval of raw_prediction is (-inf, inf) and
# we do not need to distinguish.
raw_prediction = rng.normal(loc=0, scale=10, size=(n_samples))
raw_prediction = rng.uniform(low=-20, high=20, size=(n_samples))

assert_allclose(link.link(link.inverse(raw_prediction)), raw_prediction)
y_pred = link.inverse(raw_prediction)
Expand Down