-
-
Notifications
You must be signed in to change notification settings - Fork 25.8k
[MRG+1] break the tie in Meanshift in case cluster intensities are the same #11901
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.
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
Conversation
interesting. can we add a more specific test for this? also, if we're going
to fix for release, please update what's new
|
yeah, I added a test for that case. Which is basically the example which was failing. |
Please add an entry to the change log at |
This doesn't lead to a failure right now so isn't critical, right? |
Well it was leading to a failure. Did they change? |
It's only not failing because we're not running the doctests. The test case
here would presumably fail on some platforms. So if you want it to be
critical, we can add the test case without the fix :p
|
LGTM as well, indeed it's useful to merge for 0.20 because of #12014. |
See MacPython/scikit-learn-wheels#7 (comment)
MeanShift is undeterministic if intensities of clusters are the same. This uses the cluster center values to sort in case of a tie.