Page MenuHomeSolus

Updated scikit-learn to 0.21.2
ClosedPublic

Authored by maverick1 on Tue, Jul 2, 11:52 AM.

Details

Summary

Updated scikit-learn to 0.21.

Changes:

  • Fixed a bug in cross_decomposition.CCA improving numerical stability when Y is close to zero
  • Fixed a bug in metrics.pairwise.euclidean_distances where a part of the distance matrix was left un-instanciated for suffiently large float32 datasets (regression introduced in 0.21).
  • Fixed a bug in preprocessing.OneHotEncoder where the new drop parameter was not reflected in get_feature_names.
  • Fixed an issue with plot_tree where it display entropy calculations even for gini criterion in DecisionTreeClassifiers
  • Fixed a bug where min_max_axis would fail on 32-bit systems for certain large inputs. This affects preprocessing.MaxAbsScaler, preprocessing.normalize and preprocessing.LabelBinarizer.
Test Plan

Ran examples codes from the website and plotted in matplotlib

Diff Detail

Repository
R2855 scikit-learn
Lint
Automatic diff as part of commit; lint not applicable.
Unit
Automatic diff as part of commit; unit tests not applicable.
maverick1 created this revision.Tue, Jul 2, 11:52 AM
maverick1 requested review of this revision.Tue, Jul 2, 11:52 AM
JoshStrobl accepted this revision.Tue, Jul 2, 5:12 PM
JoshStrobl edited the summary of this revision. (Show Details)
JoshStrobl added a subscriber: JoshStrobl.

LGTM, thanks!

This revision is now accepted and ready to land.Tue, Jul 2, 5:12 PM
This revision was automatically updated to reflect the committed changes.

@JoshStrobl why did you remove .2 of the summary?

@JoshStrobl why did you remove .2 of the summary?

Accidental when I went to cleanup the lack of newlines.