Skip to content
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

Jump to bottom

[MRG] Nearest Centroid Divide by Zero Fix #18370

Merged
merged 12 commits into from Oct 2, 2020
Merged

[MRG] Nearest Centroid Divide by Zero Fix #18370

Show file tree
Hide file tree
Changes from 5 commits
Commits
Show all changes
12 commits
Select commit Hold shift + click to select a range
063e801
added value error for divide by zero caused by zero variance in all f…
trewaite Sep 7, 2020
81c1bbf
formatting fix
trewaite Sep 7, 2020
36bc1fb
added test for zero div value error
trewaite Sep 24, 2020
a2b396f
fixed formatting
trewaite Sep 24, 2020
fcc93eb
fixed formatting
trewaite Sep 24, 2020
1bf3b36
using np.ptp instead of variance to find constant features
trewaite Sep 24, 2020
a1d6386
added more robust test case for test_features_zero_var
trewaite Sep 24, 2020
c4da493
fixed check fail by adding arg axis=0 to np.ptp
trewaite Sep 24, 2020
58174ee
using np.all for constant features check to avoid numerical instability
trewaite Sep 24, 2020
d713b6a
updated change log v0.24
trewaite Sep 30, 2020
d562627
Update doc/whats_new/v0.24.rst
trewaite Sep 30, 2020
83bbd42
added new username whats_new rst
trewaite Sep 30, 2020
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Jump to
Jump to file
Failed to load files.
Diff view
Diff view
3 changes: 3 additions & 0 deletions sklearn/neighbors/_nearest_centroid.py
View file
Open in desktop
Expand Up @@ -161,6 +161,9 @@ def fit(self, X, y):
# Calculate deviation using the standard deviation of centroids.
variance = (X - self.centroids_[y_ind]) ** 2
variance = variance.sum(axis=0)
if np.sum(variance) == 0:
raise ValueError("All features have zero variance. "
"Division by zero.")
Copy link
Member

Choose a reason for hiding this comment

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

We would been to be careful here with using variance directly:

import numpy as np

X = np.empty((10, 2))
X[:, 0] = -0.13725701
X[:, 1] = -0.9853293

X_means = X.mean(axis=0)
var = (X - means)**2
var = var.sum(axis=0)

var
# array([7.70371978e-33, 1.23259516e-31])

We can use np.ptp instead to look for constant features:

np.ptp(X, axis=0)
# array([0., 0.])

Copy link
Contributor Author

Choose a reason for hiding this comment

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

No problem! Good catch, changed logic to np.ptp(X).sum() == 0 and changed test case to match your example.

s = np.sqrt(variance / (n_samples - n_classes))
s += np.median(s) # To deter outliers from affecting the results.
mm = m.reshape(len(m), 1) # Reshape to allow broadcasting.
Expand Down
10 changes: 10 additions & 0 deletions sklearn/neighbors/tests/test_nearest_centroid.py
View file
Open in desktop
Expand Up @@ -146,3 +146,13 @@ def test_manhattan_metric():
clf.fit(X_csr, y)
assert_array_equal(clf.centroids_, dense_centroid)
assert_array_equal(dense_centroid, [[-1, -1], [1, 1]])


def test_features_zero_var():
# Test that features with 0 variance throw error

X = np.array([[0, 0], [0, 0], [0, 0], [0, 0], [0, 0]])
y = np.array([1, 0, 0, 1, 0])
clf = NearestCentroid(shrink_threshold=0.1)
with assert_raises(ValueError):
clf.fit(X, y)