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
[MRG] Release 0.20.4 #14443
[MRG] Release 0.20.4 #14443
Conversation
…ifiedKFold (scikit-learn#13124) * Enable StratifiedKFold to produce different splits * what's new * redundant statement * update what's new * redundant comment * add a test * move what's new entry * review comment * review comment
scikit-learn#14237) * FIX Raise error on reordered columns in ColumnTransformer with remainder * FIX Check for different length of X.columns to avoid exception * FIX linter, line too long * FIX import _check_key_type from its new location utils * ENH Adjust doc, allow added columns * Fix comment typo as suggested, remove non-essential exposition in doc * Add PR 14237 to what's new * Avoid AttributeError in favor of ValueError "column names only for DF" * ENH Add check for n_features_ for array-likes and DataFrames * Rename self.n_features to self._n_features * Replaced backslash line continuation with parenthesis * Style changes
AppVeyor does not appear to be running. Have we disabled it? |
Conda is throwing PackageNotFoundErrors that I'm not sure how to resolve. |
The last commit has strangely not appeared here. |
* [scipy-dev] manually scramble the indices * [scipy-dev] review comment
Codecov Report
@@ Coverage Diff @@
## 0.20.X #14443 +/- ##
==========================================
- Coverage 93.13% 91.59% -1.55%
==========================================
Files 433 423 -10
Lines 75144 73552 -1592
==========================================
- Hits 69987 67367 -2620
- Misses 5157 6185 +1028
Continue to review full report at Codecov.
|
Finally it's green! |
Neither has until now been proposed for a patch release as far as I know, not for 0.20 nor for 0.21 |
Could core devs please review and approve release? Thanks |
Should we fix, or open an issue for some the commits you have here, like 7b0b97b (sorry if that's done and I've missed them)? |
I am pretty sure they are covered elsewhere. Some things like warnings in tests have quick fixes in the patch branch because their fix in master was more substantial or depended on other changes. |
@adrinjalali, for that specific issue see #14125 |
Are you fine with this one too, @qinhanmin2014? |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
CIs are green and the diff seems reasonable, so LGTM and thanks for making this happen.
Though personally I'm not a big fan of 0.20.X, backporting is time-consuming and I guess it's impossible to backport all the bug fixes (only backporting selected bug fixes seems strange). If users want to avoid bugs, they should turn to python3.X.
.............................. | ||
|
||
- |Fix| Fixed a bug in :class:`cluster.KMeans` where KMeans++ initialisation | ||
could rarely result in an IndexError. :issue:`11756` by `Joel Nothman`_. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
this entry is not included in master (and 0.21.3)?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Ahh, true. It's now included in the 0.20.4 section in master at least...
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Sorry for missing it.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Also, would be happy to give a hand on the task list you have on each release.
The task list is not a great effort... I have been wondering though about
whether we can put it all into an Azure deployment Pipeline (cf. #13949).
It would help by only actually doing the tagging if the wheels (and ideally
conda-forge) are successfully built, and it would avoid running back and
forth between here and macpython.
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks for doing the release work @jnothman ! The list of commits and the release notes LGTM.
Building wheels at https://travis-ci.org/MacPython/scikit-learn-wheels/builds/564977472 |
Thanks. In case you missed some emails, my review is silently ignored (#14443 (comment)), but that's trivial. |
Not to be squashed.
This takes the PRs from https://github.com/scikit-learn/scikit-learn/milestone/30?closed=1, constructs a change log for them, upgrades vendored Joblib from 0.13.0 to 0.13.2, proposes a release date of 29 July (the coming Monday) and bumps the version.
I would aim to synchronise release with 0.21.3 (#14188), but this one is good to go (pending CI and review) as far as I'm concerned.
TODO after approval:
(regenerate Dash docs: https://github.com/Kapeli/Dash-User-Contributions/tree/master/docsets/Scikit)