Release 0.21.2 #13915

jnothman · 2019-05-21T08:19:34Z

Patch release for the euclidean distances fix. We can delay this if we want, but it's a pretty high risk issue.

Rebase so far looks like:

d 0b3aa5e466 DOC bump version
d 10249939e5 DOC move 0.20 to previous releases
d e7bd8a33e5 FIX Optics paper typo which resulted in undersized clusters (#13750)
d 384c8ad3d3 ENH add a break_ties parameter to SVC and NuSVC (#12557)
d 13981bdce9 STY Remove variable renaming (#13731)
d 19192c0427 MAINT Replace manual checks with `check_is_fitted`  (#13013)
d 905dedc635 DEP remove random_state from OneClassSVM (#13802)
d 28480f4cb2 DEP change the default of gamma in SVM (#13801)
d d0a94fea74 DEP remove graph lasso (#13795)
d 7896b21562 MNT change default of solver in LogisiticRegression (#13805)
d 6525a39dd6 MNT Remove Imputer in preprocessing (#13796)
d 43f85020c5 DEP remove correlation and regression models from GaussianProcess (#13819)
d 2b7a69baca DEP remove the batch_size parameter from pariwise_distance_argmin (#13822)
d 53624e8c26 DEP remove public function download_20newsgroups (#13829)
d 9631a67388 DEP remove scale_face function from lfw (#13830)
d 777f91f114 [MRG] DEP Removed the reorder parameter from the auc function (#13827)
d 7166ae8ece MNT Updated adjusted_mutual_info_score and normalized_mutual_info_score default to 'arithmetic' (#13814)
d 91e019df5e DEP remove precomputed parameter in t_sne.trustworthiness (#13820)
d 69eb4d4678 [MRG] DEP change default and deprecate iid in SearchCV (#13834)
d c7566ea232 disallow input as sparse matrix in affinity propagation function, Aff… (#13828)
d 95339a677e [MRG] DEP remove pooling_func in AgglomerativeClustering (#13821)
d 7fdac52c19 MNT Remove backward compatibility of param order in make_column_transformer (#13831)
d 8a8e21b2a3 MNT Change the default value of n_estimators in forests (#13803)
d 83656484a8 MNT Remove raises and with_setup requiring nose (#13842)
d da9e680996 DOC fix the docstring of AMI MNI regarding new default (#13837)
d eec7614a1f Small fixes to maintainer commands
d b243c6aa43 DEP change default of contamination in LOF (#13815)
d 75967ef20e DEP remove deprecated parameters in EllipticEnvelope (#13818)
p 17eb3c9ca0 DOC Update README.rst (#13879)
p e76984b0a1 [MRG] doc about creating an experimental module (#13865)
p c890ecd923 DOC Fix missing newline in DistanceMetric docs (#13888)
d 11de9d60f2 Updating MissingIndicator User Guide section (#13849)
d 4e65827b5c [MRG] RidgeCV minor refactor to improve readability (#13832)
d 5d240c6a0b MAINT Fixes coverage reporting on pylatest_conda (#13895)
p 2fd1a6725b DOC Improve documentation of R-squared in User Guide (#13867)
p ea58e27ace DOC Documentation improvement in fbeta_score and OneHotEncoder (#13904)
p eb93420e87 DOC Rewrite algorithm selection section for nearest neighbours (#13906)
p 98aefc1fdc FIX regression in eulidean distances (float32) related to batch management (#13910)
p 0a2dc72982 TST improve warning reporting (#12597)
d af4247b152 DEP remove utilities related to mldata (#13798)
d 0bfa52dad7 DEP Change default of error_score in cross-validation (#13840)
d 911792b008 ENH Avoid calling _encode_check_unknown() twice in BaseEncoder.transform (#13810)
d 57c04f4e8d ENH Allows setting of initial hyperparameters for BayesianRidge (#13618)
d 778b11904e [MRG] DEP remove threshold_, change contamination, deprecate behaviour in iforest (#13811)
d c28ef9ef2a DEP change the default of multi_class to 'auto' in logistic regression (#13807)
p 62a0bcdbe1 Better fix the rng seed in test_fastica_simple (#13848)
p c418761265 FIX clip small values in PLS cross-decomposition for increased stability (#13903)
d e0532cdea5 MAINT increase numerical gradient check tolerance to make the test stable (#13885)
p e35f040200 [MRG] Fixes get_feature_names results when using drop functionality (#13894)
p dc5e4d8896 FIX downcast large matrix indices where possible in sparsefuncs._minor_reduce (fix #13737) (#13741)

)

…t-learn#13904)

…t-learn#13906)

…ement (scikit-learn#13910)

jnothman · 2019-05-21T10:20:11Z

The windows failure is cross_decomposition's test_scale_and_stability

jnothman · 2019-05-21T10:21:04Z

The windows failure is cross_decomposition's test_scale_and_stability. Not checked if this is a known issue, but it obviously did not fail at the previous patch release

jnothman · 2019-05-21T10:22:51Z

I see #13903 now

ogrisel · 2019-05-22T13:04:54Z

+1 for releasing as soon as #13903 is in.

…ity (scikit-learn#13903)

ogrisel · 2019-05-23T10:57:07Z

I relaunched the azure build because there was an HTTP error in the ubuntu build.

jnothman · 2019-05-23T11:28:31Z

Yeah, that build hasn't been working all day: #13928

jnothman · 2019-05-23T11:30:13Z

So I'll probably just do the release despite that build. But I was still hoping to throw in #13894... Would you mind looking at it @ogrisel?

jnothman · 2019-05-23T11:54:47Z

#13741 could also be shipped if we want

thomasjpfan · 2019-05-23T13:04:37Z

test_scale_and_stability may still be a little flaky: #13930

…cikit-learn#13894)

…r_reduce (fix scikit-learn#13737) (scikit-learn#13741)

jnothman · 2019-05-23T13:17:38Z

@thomasjpfan you're good with me merging and releasing if this turns green except for Linux py35_np_atlas?

jnothman · 2019-05-23T14:07:04Z

The todo list once we have consensus to release:

merge branch
create tag
update news and what's new date in master and release branch
update dependencies and release tag at https://github.com/MacPython/scikit-learn-wheels
twine the wheels to PyPI when that's green
https://github.com/scikit-learn/scikit-learn/releases draft
confirm bot detected at https://github.com/conda-forge/scikit-learn-feedstock and wait for merge
https://github.com/scikit-learn/scikit-learn/releases publish
announce on mailing list

thomasjpfan · 2019-05-23T14:36:50Z

Yea I am okay with this release besides the py35_np_atlas test. I am seeing if there is a way to work around it. I want to see if test_scale_and_stability works on that instance.

jnothman · 2019-05-23T14:48:49Z

But it doesn't exactly harm the release by excluding that patch? Do you have any idea why those packages may have disappeared?

jnothman · 2019-05-23T14:57:10Z

I'm going to go ahead with this now, largely because it suits me schedule...

thomasjpfan · 2019-05-23T15:00:25Z

Okay this works for me. I opened #13934 to deal with the apt issue and to make sure our tests pass on py35_np_atlas.

ogrisel · 2019-05-23T17:49:45Z

I really don't understand the macOS / Python 3.5 failure on the wheel builder CI:

https://travis-ci.org/MacPython/scikit-learn-wheels/jobs/536336401

...
Requirement already satisfied: numpy>=1.11.0 in ./venv/lib/python3.5/site-packages (from scikit-learn==0.21.2) (1.11.0)
Installing collected packages: scikit-learn
Successfully installed scikit-learn-0.21.2
Python 3.5.4
============================= test session starts ==============================
platform darwin -- Python 3.5.4, pytest-4.5.0, py-1.8.0, pluggy-0.11.0
rootdir: /Users/travis/build/MacPython/scikit-learn-wheels/tmp_for_test
collected 0 items / 1 errors
==================================== ERRORS ====================================
________________________ ERROR collecting test session _________________________
../venv/lib/python3.5/site-packages/_pytest/config/__init__.py:437: in _importconftest
    return self._conftestpath2mod[conftestpath]
E   KeyError: local('/Users/travis/build/MacPython/scikit-learn-wheels/venv/lib/python3.5/site-packages/sklearn/conftest.py')
During handling of the above exception, another exception occurred:
../venv/lib/python3.5/site-packages/_pytest/config/__init__.py:443: in _importconftest
    mod = conftestpath.pyimport()
../venv/lib/python3.5/site-packages/py/_path/local.py:701: in pyimport
    __import__(modname)
../venv/lib/python3.5/site-packages/sklearn/__init__.py:76: in <module>
    from .base import clone
../venv/lib/python3.5/site-packages/sklearn/base.py:16: in <module>
    from .utils import _IS_32BIT
../venv/lib/python3.5/site-packages/sklearn/utils/__init__.py:13: in <module>
    from scipy.sparse import issparse
../venv/lib/python3.5/site-packages/scipy/sparse/__init__.py:230: in <module>
    from .csr import *
../venv/lib/python3.5/site-packages/scipy/sparse/csr.py:13: in <module>
    from ._sparsetools import (csr_tocsc, csr_tobsr, csr_count_blocks,
E   ImportError: numpy.core.multiarray failed to import
During handling of the above exception, another exception occurred:
../venv/lib/python3.5/site-packages/_pytest/runner.py:225: in from_call
    result = func()
../venv/lib/python3.5/site-packages/_pytest/runner.py:252: in <lambda>
    call = CallInfo.from_call(lambda: list(collector.collect()), "collect")
../venv/lib/python3.5/site-packages/_pytest/main.py:542: in collect
    for x in self._collect(arg):
../venv/lib/python3.5/site-packages/_pytest/main.py:570: in _collect
    col = self._collectfile(pkginit, handle_dupes=False)
../venv/lib/python3.5/site-packages/_pytest/main.py:635: in _collectfile
    ihook = self.gethookproxy(path)
../venv/lib/python3.5/site-packages/_pytest/main.py:481: in gethookproxy
    my_conftestmodules = pm._getconftestmodules(fspath)
../venv/lib/python3.5/site-packages/_pytest/config/__init__.py:421: in _getconftestmodules
    mod = self._importconftest(conftestpath.realpath())
../venv/lib/python3.5/site-packages/_pytest/config/__init__.py:460: in _importconftest
    raise ConftestImportFailure(conftestpath, sys.exc_info())
E   _pytest.config.ConftestImportFailure: (local('/Users/travis/build/MacPython/scikit-learn-wheels/venv/lib/python3.5/site-packages/sklearn/conftest.py'), (<class 'ImportError'>, ImportError('numpy.core.multiarray failed to import',), <traceback object at 0x105809d08>))

numpy is missing from pytest but pip says "Requirement already satisfied: numpy>=1.11.0".

ogrisel · 2019-05-23T21:06:48Z

There is also the following warning:

/Users/travis/build/MacPython/scikit-learn-wheels/venv/lib/python3.5/site-packages/scipy/__init__.py:115
  /Users/travis/build/MacPython/scikit-learn-wheels/venv/lib/python3.5/site-packages/scipy/__init__.py:115: UserWarning: Numpy 1.13.3 or above is required for this version of scipy (detected version 1.11.0)
    UserWarning)

Let me try to do a PR.

jnothman · 2019-05-23T21:34:57Z

It's amazing how much can change from week to week...

willdarnell and others added 10 commits May 21, 2019 18:13

DOC Update README.rst (scikit-learn#13879)

85828d6

[MRG] doc about creating an experimental module (scikit-learn#13865)

1dc0898

DOC Fix missing newline in DistanceMetric docs (scikit-learn#13888)

c2ce254

DOC Improve documentation of R-squared in User Guide (scikit-learn#13867

e6a45f9

)

DOC Documentation improvement in fbeta_score and OneHotEncoder (sciki…

adf70f4

…t-learn#13904)

DOC Rewrite algorithm selection section for nearest neighbours (sciki…

d14400e

…t-learn#13906)

FIX regression in eulidean distances (float32) related to batch manag…

ef9f42e

…ement (scikit-learn#13910)

TST improve warning reporting (scikit-learn#12597)

10050b6

MNT bump version

6d19913

DOC set release dates

6ed7e6a

rth mentioned this pull request May 21, 2019

Run check_manifest in one of the CI setups #13916

Closed

ogrisel and others added 2 commits May 23, 2019 12:31

Better fix the rng seed in test_fastica_simple (scikit-learn#13848)

3b5ebe0

FIX clip small values in PLS cross-decomposition for increased stabil…

b1a531a

…ity (scikit-learn#13903)

jamesmyatt and others added 3 commits May 23, 2019 23:12

[MRG] Fixes get_feature_names results when using drop functionality (s…

45ab4c6

…cikit-learn#13894)

FIX downcast large matrix indices where possible in sparsefuncs._mino…

e833c92

…r_reduce (fix scikit-learn#13737) (scikit-learn#13741)

TST Sets decimal

e8602bc

jnothman merged commit e8602bc into scikit-learn:0.21.X May 23, 2019

ogrisel mentioned this pull request May 23, 2019

Fix macos version matrix MacPython/scikit-learn-wheels#29

Merged

Uh oh!

Release 0.21.2 #13915

Release 0.21.2 #13915

Uh oh!

Conversation

jnothman commented May 21, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

jnothman commented May 21, 2019

Uh oh!

jnothman commented May 21, 2019

Uh oh!

jnothman commented May 21, 2019

Uh oh!

ogrisel commented May 22, 2019

Uh oh!

ogrisel commented May 23, 2019

Uh oh!

jnothman commented May 23, 2019

Uh oh!

jnothman commented May 23, 2019

Uh oh!

jnothman commented May 23, 2019 via email

Uh oh!

thomasjpfan commented May 23, 2019

Uh oh!

jnothman commented May 23, 2019

Uh oh!

jnothman commented May 23, 2019 • edited by ogrisel Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

thomasjpfan commented May 23, 2019

Uh oh!

jnothman commented May 23, 2019 via email

Uh oh!

jnothman commented May 23, 2019 via email

Uh oh!

thomasjpfan commented May 23, 2019

Uh oh!

ogrisel commented May 23, 2019

Uh oh!

ogrisel commented May 23, 2019

Uh oh!

jnothman commented May 23, 2019 via email

Uh oh!

Uh oh!

jnothman commented May 21, 2019 •

edited

Loading

jnothman commented May 23, 2019 •

edited by ogrisel

Loading