8000 DOC Updated documentation for cv parameter (issue #4533) by christophebourguignat · Pull Request #5184 · scikit-learn/scikit-learn · GitHub
[go: up one dir, main page]
Skip to content

DOC Updated documentation for cv parameter (issue #4533)#5184

Merged
GaelVaroquaux merged 1 commit intoscikit-learn:masterfrom
christophebourguignat:master
Aug 30, 2015
Merged

DOC Updated documentation for cv parameter (issue #4533)#5184
GaelVaroquaux merged 1 commit intoscikit-learn:masterfrom
christophebourguignat:master

Conversation

@christophebourguignat
Copy link
@christophebourguignat christophebourguignat commented Aug 29, 2015

No description provided.

@jnothman
Copy link
Member
jnothman commented Aug 30, 2015

Thanks. This is applicable everywhere there is a cv parameter and a classifier can be passed. I suspect @rvraghav93 is right that we should keep it brief and point to centralised documentation, perhaps just the documentation of check_cv (once clarified, if necessary).

@FedericoV
Copy link
Contributor
FedericoV commented Aug 30, 2015

This is a good improvement over what was there before. The other places where it shows up are:

666 : 1063 : def cross_val_predict(estimator, X, y=None, cv=None, n_jobs=1,
675 : 1225 : def cross_val_score(estimator, X, y=None, scoring=None, cv=None, n_jobs=1,
696 : 1543 : def permutation_test_score(estimator, X, y, cv=None,
in cross_validation.py

and

2993 : 232 : def validation_curve(estimator, X, y, param_name, param_range, cv=None,
in validation_curve.py

@GaelVaroquaux
Copy link
Member
GaelVaroquaux commented Aug 30, 2015

This is useful. Thanks! Merging.

GaelVaroquaux added a commit that referenced this pull request Aug 30, 2015
@GaelVaroquaux
Merge pull request #5184 from christophebourguignat/master
869971f 
DOC Updated documentation for cv parameter (issue #4533)
@GaelVaroquaux GaelVaroquaux merged commit 869971f into scikit-learn:master Aug 30, 2015
@jnothman
Copy link
Member
jnothman commented Aug 31, 2015

Someone want to copy to those other destinations?

@christophebourguignat
Copy link
Author
christophebourguignat commented Aug 31, 2015

@jnothman I can do it

@christophebourguignat
Copy link
Author
christophebourguignat commented Aug 31, 2015

Just proposed a new PR.

A few questions :

  • in function template we say default cv is None, but in documentation default cv is 3. In check_cv() code this is the same (if cv is None: cv = 3), but shall we unify in the doc (or the code) ?
  • shall we also update cv doc in sklearn.calibration.CalibratedClassifierCV ?

@amueller
Copy link
Member
amueller commented Sep 8, 2015

@christophebourguignat which one was your PR?
The main reason for not doing cv = 3 in all the constructors is that we don't want to hard-code that in so many places iirc. Currently the number 3 is only written down in check_cv.
But we don't want our users to have to dig until they find check_cv to find out what that means.
We could change the default in the docs to None if the rest of the docstring is informative enough.

@raghavrv
Copy link
Member
raghavrv commented Sep 10, 2015

Thanks for the PR this will add more consistency to the cv param documentation.

But @jnothman @GaelVaroquaux @amueller didn't we agree upon cv doc as follows? -

    cv : int, cross-validation generator or an iterable, optional
        Determines the cross-validation splitting strategy.
        Possible inputs for cv are:
          - None, to use the default 3-fold cross-validation,
          - integer, to specify the number of folds.
          - An object to be used as a cross-validation generator.
          - An iterable yielding train/test splits.

        For integer/None inputs, if ``y`` is binary or multiclass,
        :class:`StratifiedKFold` used. If classifier is False or if ``y`` is
        neither binary nor multiclass, :class:`KFold` is used.

        Refer :ref:`User Guide <cross_validation>` for the various
        cross-validation strategies that can be used here.

The current version is -

     cv : integer or cross-validation generator, default=3
        If an integer is passed, it is the number of folds.
        A cross-validation generator to use. If int, determines
        Specific cross-validation objects can be passed, see
        the number of folds in StratifiedKFold if estimator is a classifier
        sklearn.cross_validation module for the list of possible objects
        and the target y is binary or multiclass, or the number
        of folds in KFold otherwise.
        Specific cross-validation objects can be passed, see 
        sklearn.cross_validation module for the list of possible objects.

@raghavrv
Copy link
Member
raghavrv commented Sep 10, 2015

If you are okay with the first doc, let me know I'll add a quick PR and merge the change to #4294 too!

@amueller
Copy link
Member
amueller commented Sep 10, 2015

It was a net improvement. The first one looks nicer.

@raghavrv raghavrv mentioned this pull request Sep 10, 2015
Merged
@raghavrv
Copy link
Member
raghavrv commented Sep 10, 2015

#5238

@raghavrv raghavrv mentioned this pull request Sep 10, 2015
Merged
24 tasks
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

6 participants

@christophebourguignat @jnothman @FedericoV @GaelVaroquaux @amueller @raghavrv
0