Codestin Search App

equialgo · 2015-11-02T23:57:53Z

… when using Pandas DataFrame/Series

Related to issue #5696

raghavrv · 2015-11-30T23:56:44Z

thanks for fixing this!

Could you also add this fix to the model_selection module and also add a NRT to model_selection/tests/test_validation.py? :)

raghavrv · 2015-11-30T23:58:39Z

Also could you edit the

PR title to -
[MRG] split data using _safe_split in _permutaion_test_scorer to fix error when using Pandas DataFrame/Series

And the PR description to "Fixes #5696"

amueller · 2015-12-10T17:23:31Z

yeah a test would be great, using the mock data frame or an acutal dataframe if pandas is installed.

equialgo · 2016-01-03T17:06:25Z

First of all, sorry for the late reply.

I added the fix also to model_selection/_validation.py.

Moreover, I added a test_permutation_test_score_pandas() test case to model_selection/tests/test_validation.py (based on the test_cross_val_score_pandas() test case)

dankessler · 2016-12-14T15:19:16Z

Is there additional work that needs to be done before this can be merged? I'm still encountering the issue that this addresses (#5696), which I'm working around by calling .values on my series. Is it just the merge conflict that's hairy? If so, I'm happy to take a crack.

amueller · 2016-12-14T16:34:14Z

Wow sorry this one has been lying around for a bit. It needs a rebase and an entry to whatsnew.rst - and reviews, though that should be reasonably quick.

amueller · 2016-12-14T16:36:59Z

This actually fixes another issue, namely using precomputed kernels / distance matrices with permutation_score. It would be great if there was also a regression test for that.
Otherwise looks good to me. Sorry for the slow turn-around.

jnothman · 2016-12-14T23:02:15Z

sklearn/model_selection/tests/test_validation.py

+        check_df = lambda x: isinstance(x, InputFeatureType)
+        check_series = lambda x: isinstance(x, TargetType)
+        clf = CheckingClassifier(check_X=check_df, check_y=check_series)
+        permutation_test_score(clf, X_df, y_ser)


Ideally we should check that, for fixed random_state, results are identical regardless of input type.

At the moment, such input tests are also not present for the other validation methods, there is only: test_cross_val_score_pandas(), test_cross_val_predict_pandas().

I wonder is that really a necessity?

jnothman · 2016-12-14T23:02:55Z

@equialgo Sorry for the extreme delay. Let us know if you have time soon to complete the changes.

equialgo · 2016-12-24T14:46:51Z

Hi guys, I rebased the fix it should be relatively easy to merge now (if it passes the checks).

@amueller I do not really get why this also fixes the issue using precomputed kernels / distance matrices with permutation_score. If it is trivial please elaborate, otherwise it might be better to add a new ticket for that one.

jnothman · 2016-12-25T10:15:50Z

You've done something strange with your commit history. Perhaps a rebase is in order.

equialgo · 2016-12-28T16:35:54Z

@jnothman oops, kinda messed-up there! Re-did the rebase; commit history should be okay now.

…hed commits) Squashed commits: [94fd9f4] split data using _safe_split in _permutaion_test_scorer [522053b] adding test case test_permutation_test_score_pandas() to check if permutation_test_score plays nice with pandas dataframe/series [21b23ce] running test_permutation_test_score_pandas on iris data to prevent warnings. [15a48bf] adding safe_indexing to _shuffle function [9ea5c9e] adding test case test_permutation_test_score_pandas() to check if permutation_test_score plays nice with pandas dataframe/series [3cf5e8f] split data using _safe_split in _permutaion_test_scorer to fix error when using Pandas DataFrame/Series

GaelVaroquaux · 2016-12-28T20:44:04Z

+1 to merge this guy (there is a flake8 failure, but it's because of lambda expressions, and here they seem legit).

Let's try to merge this fast, it's been lying around for too long.

GaelVaroquaux · 2016-12-28T20:46:51Z

@raghavrv : are you +1 for merge? If so, let's merge

raghavrv · 2016-12-28T20:47:59Z

I think this needs a whatsnew entry? I can send a quick PR after this gets merged if you want. Otherwise LGTM... That pep8 error needs to be added to ignore list... (#8131)

raghavrv

Pending whatsnew... (Feel free to merge and later add a whatsnew...)

raghavrv · 2016-12-29T01:48:11Z

Okay... I'm merging this... Il make a whatsnew entry for this and another PR tomorrow...

raghavrv · 2016-12-29T01:49:34Z

Thanks @equialgo!

equialgo · 2016-12-29T08:26:12Z

You guys thanks for merging!

…arn#5697) Squashed commits: [94fd9f4] split data using _safe_split in _permutaion_test_scorer [522053b] adding test case test_permutation_test_score_pandas() to check if permutation_test_score plays nice with pandas dataframe/series [21b23ce] running test_permutation_test_score_pandas on iris data to prevent warnings. [15a48bf] adding safe_indexing to _shuffle function [9ea5c9e] adding test case test_permutation_test_score_pandas() to check if permutation_test_score plays nice with pandas dataframe/series [3cf5e8f] split data using _safe_split in _permutaion_test_scorer to fix error when using Pandas DataFrame/Series

raghavrv mentioned this pull request Nov 30, 2015

Using a Pandas Series/DataFrame with permutation_test_score results in error. #5696

Closed

amueller added the Waiting for Reviewer label Dec 10, 2015

amueller changed the title ~~split data using _safe_split in _permutaion_test_scorer to fix error…~~ [MRG] split data using _safe_split in _permutaion_test_scorer to fix error… Dec 10, 2015

amueller added the Bug label Dec 14, 2016

jnothman reviewed Dec 14, 2016

View reviewed changes

equialgo force-pushed the fix-issue-5696 branch from 210d663 to f0bc2cd Compare December 28, 2016 16:29

equialgo force-pushed the fix-issue-5696 branch from f0bc2cd to de14dfa Compare December 28, 2016 17:01

GaelVaroquaux changed the title ~~[MRG] split data using _safe_split in _permutaion_test_scorer to fix error…~~ [MRG+1] split data using _safe_split in _permutaion_test_scorer to fix error… Dec 28, 2016

raghavrv approved these changes Dec 28, 2016

View reviewed changes

raghavrv removed the Waiting for Reviewer label Dec 28, 2016

raghavrv merged commit 986a49b into scikit-learn:master Dec 29, 2016

Przemo10 mentioned this pull request Mar 17, 2017

update fork (#1) #8606

Closed

Uh oh!

Conversation

equialgo commented Nov 2, 2015

Uh oh!

raghavrv commented Nov 30, 2015

Uh oh!

raghavrv commented Nov 30, 2015

Uh oh!

amueller commented Dec 10, 2015

Uh oh!

equialgo commented Jan 3, 2016

Uh oh!

dankessler commented Dec 14, 2016

Uh oh!

amueller commented Dec 14, 2016

Uh oh!

amueller commented Dec 14, 2016

Uh oh!

jnothman Dec 14, 2016

Choose a reason for hiding this comment

Uh oh!

equialgo Dec 24, 2016

Choose a reason for hiding this comment

Uh oh!

jnothman commented Dec 14, 2016

Uh oh!

equialgo commented Dec 24, 2016 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

jnothman commented Dec 25, 2016

Uh oh!

equialgo commented Dec 28, 2016

Uh oh!

GaelVaroquaux commented Dec 28, 2016

Uh oh!

GaelVaroquaux commented Dec 28, 2016

Uh oh!

raghavrv commented Dec 28, 2016 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

raghavrv left a comment

Choose a reason for hiding this comment

Uh oh!

raghavrv commented Dec 29, 2016 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

raghavrv commented Dec 29, 2016

Uh oh!

equialgo commented Dec 29, 2016

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

Comments

equialgo commented Dec 24, 2016 •

edited

Loading

raghavrv commented Dec 28, 2016 •

edited

Loading

raghavrv commented Dec 29, 2016 •

edited

Loading