Skip to content

[MRG] Fix GaussianMixture fit_predict != fit.predict when n_init > 1 - #13142

Merged
ogrisel merged 3 commits into
scikit-learn:masterfrom
jeremiedbb:fix-gaussian-mixture-fit-predict
Feb 25, 2019
Merged

ogrisel merged 3 commits into
scikit-learn:masterfrom
jeremiedbb:fix-gaussian-mixture-fit-predict

Conversation

@jeremiedbb

Copy link
Copy Markdown
Member

Fixes #13070

Move the last _e_step after having set the params to the best one found.
Added a test, which fails on master.

@jeremiedbb

Copy link
Copy Markdown
Member Author

Does it go into 0.20.3 or into 0.21 ?

@jnothman jnothman left a comment

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

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

We would ordinarily release this in 0.21, but I don't know whether we want to make 0.20.3 more extensive as an LTS and hence whether the recent patch to Stratified gold belongs there too.

@jnothman

Copy link
Copy Markdown
Member

Please add what's new

@jeremiedbb

Copy link
Copy Markdown
Member Author

Since the issue was in BaseMixture it also affects BayesianGaussianMixture and indeed the same test fails on master with this estimator. So I added this test in test_bayesian_mixture as well.

@ogrisel
ogrisel merged commit 3a47124 into scikit-learn:master Feb 25, 2019
@ogrisel

ogrisel commented Feb 25, 2019 •

Copy link
Copy Markdown
Member

Merged. Thanks @jeremiedbb!

xhluca pushed a commit to xhluca/scikit-learn that referenced this pull request Apr 28, 2019
xhluca pushed a commit to xhluca/scikit-learn that referenced this pull request Apr 28, 2019
koenvandevelde pushed a commit to koenvandevelde/scikit-learn that referenced this pull request Jul 12, 2019
@jeremiedbb
jeremiedbb deleted the fix-gaussian-mixture-fit-predict branch September 3, 2026 16:10
davidgegiya added a commit to davidgegiya/agent-orchestrator that referenced this pull request Nov 9, 2026
…gree when n_init>1

Pin 1c8668b0a021832386470ddf740d834e02c66f69 and its regression selectors for reproducible runs.

Source: scikit-learn/scikit-learn#13142
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants