Thanks to visit codestin.com
Credit goes to github.com

Skip to content

FIX always scale continuous features to unit variance in mutual info #24747

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
jeremiedbb merged 8 commits into from
Nov 16, 2022
Merged

FIX always scale continuous features to unit variance in mutual info #24747

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
2c603ae
FIX always scale continuous features to unit variance in mutual info
glemaitre Oct 24, 2022
31d200b
DOC update pr number
glemaitre Oct 24, 2022
8fc6824
Apply suggestions from code review
glemaitre Oct 25, 2022
8d18d3a
address Tim comments
glemaitre Oct 25, 2022
0965bc2
Update doc/whats_new/v1.2.rst
glemaitre Oct 26, 2022
0037d2a
address review comments
jeremiedbb Nov 16, 2022
8936b6f
Merge remote-tracking branch 'upstream/main' into pr/glemaitre/24747
jeremiedbb Nov 16, 2022
2fb8d62
trigger ci ?
jeremiedbb Nov 16, 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
6 changes: 6 additions & 0 deletions doc/whats_new/v1.2.rst
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -354,6 +354,12 @@ Changelog
:mod:`sklearn.feature_selection`
................................

- |Fix| Fix a bug in :func:`feature_selection.mutual_info_regression` and
:func:`feature_selction.mutual_info_classif`, where the continuous features
in `X` should be scaled to a unit variance independently if the target `y` is
continuous or discrete.
:pr:`24747` by :user:`Guillaume Lemaitre <glemaitre>`

:mod:`sklearn.gaussian_process`
...............................

Expand Down
7 changes: 3 additions & 4 deletions sklearn/feature_selection/_mutual_info.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -280,10 +280,9 @@ def _estimate_mi(
if copy:
X = X.copy()

if not discrete_target:
X[:, continuous_mask] = scale(
X[:, continuous_mask], with_mean=False, copy=False
)
X[:, continuous_mask] = scale(
X[:, continuous_mask], with_mean=False, copy=False
)

# Add small noise to continuous features as advised in Kraskov et. al.
X = X.astype(np.float64, copy=False)
Expand Down
29 changes: 29 additions & 0 deletions sklearn/feature_selection/tests/test_mutual_info.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -207,3 +207,32 @@ def test_mutual_info_options(global_dtype):
assert_allclose(mi_5, mi_6)

assert not np.allclose(mi_1, mi_3)


@pytest.mark.parametrize("correlated", [True, False])
def test_mutual_information_symmetry_classif_regression(correlated, global_random_seed):
"""Check that `mutual_info_classif` and `mutual_info_regression` are
symmetric by switching the target `y` as `feature` in `X` and vice
versa.

Non-regression test for:
https://github.com/scikit-learn/scikit-learn/issues/23720
"""
rng = np.random.RandomState(global_random_seed)
n = 100
d = rng.randint(10, size=n)

if correlated:
c = d.astype(np.float64)
else:
c = rng.normal(0, 1, size=n)

mi_classif = mutual_info_classif(
c[:, None], d, discrete_features=[False], random_state=global_random_seed
)

mi_regression = mutual_info_regression(
d[:, None], c, discrete_features=[True], random_state=global_random_seed
)

assert mi_classif == pytest.approx(mi_regression)