Skip to content

Prevent division by zero in GPR when y_train is constant #18388

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

Closed
wants to merge 9 commits into from
Closed

Prevent division by zero in GPR when y_train is constant #18388

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
4b0d894
add very small number to the y_train.std to avoid a divide by zero error
Sep 13, 2020
2bc2c92
add very small number to the y_train.std to avoid a divide by zero er…
Sep 13, 2020
7b91bc6
adding an if check to add epsilon only when _y_train_std==0
Sep 14, 2020
b0d2115
proper if check included to add epsilon only when _y_train_std==0
Sep 14, 2020
84a78f4
assign _y_train.std to 1 when _y_train.std is zero to avoid divide by…
Sep 19, 2020
23b06b9
suggestions, by alfaro96, included to better understanding of the cod…
Sep 20, 2020
fc332cd
minor fix, includ missing whitespace around operator
Sep 20, 2020
ea55e6a
Included test for constant target to check the private _y_train_std a…
Sep 20, 2020
63ca232
Minor fixes after checking with flake8
Sep 20, 2020
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
9 changes: 7 additions & 2 deletions sklearn/gaussian_process/_gpr.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -196,10 +196,15 @@ def fit(self, X, y):
if self.normalize_y:
self._y_train_mean = np.mean(y, axis=0)
self._y_train_std = np.std(y, axis=0)

# Assign a standard deviation of one to constant
# targets for avoiding division by zero errors
if np.iterable(self._y_train_std):
self._y_train_std[self._y_train_std == 0] = 1
else:
self._y_train_std = (
self._y_train_std if self._y_train_std != 0 else 1)
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Could you please use as done _handle_zeros_in_scale in #19361 instead?

# Remove mean and make unit variance
y = (y - self._y_train_mean) / self._y_train_std

else:
self._y_train_mean = np.zeros(1)
self._y_train_std = 1
Expand Down
14 changes: 14 additions & 0 deletions sklearn/gaussian_process/tests/test_gpr.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -536,3 +536,17 @@ def test_bound_check_fixed_hyperparameter():
periodicity_bounds="fixed") # seasonal component
kernel = k1 + k2
GaussianProcessRegressor(kernel=kernel).fit(X, y)


@pytest.mark.parametrize('kernel', kernels)
def test_constant_target(kernel):
# Test for constant target(s), the private _y_train_std attribute has
# a standard deviation of 1 instead of 0
gpr = GaussianProcessRegressor(kernel=kernel, normalize_y=True)
# create a constant targets of 2
y = np.full((1, X.shape[0]), 2).ravel()
expected_std = np.full((1, X.shape[0]), 1).ravel()
gpr.fit(X, y)
_y_train_std = gpr._y_train_std

assert_array_equal(_y_train_std, expected_std)
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Please expand the test and call predict on the fitted model as done in #19361 and give credit to @sobkevich in the changelog.