[MRG] 0.20 deprecations #9570

amueller · 2017-08-16T21:49:54Z

I haven't removed the stuff that's model_selection now, in case we want to give that another version.
Also some other stuff still needs to be done.

Remove code tagged to be removed in v.0.20. Stole todo from #10094

Things to remove:

Classes (reported here)
- cross_validation.KFold
- cross_validation.LabelKFold
- cross_validation.LeaveOneLabelOut
- cross_validation.LeaveOneOut
- cross_validation.LeavePOut
- cross_validation.LeavePLabelOut
- cross_validation.LabelShuffleSplit
- cross_validation.ShuffleSplit
- cross_validation.StratifiedKFold
- cross_validation.StratifiedShuffleSplit
- cross_validation.PredefinedSplit
- decomposition.RandomizedPCA
- gaussian_process.GaussianProcess
- grid_search.ParameterGrid
- grid_search.ParameterSampler
- grid_search.GridSearchCV
- grid_search.RandomizedSearchCV
- mixture.DPGMM
- mixture.GMM
- mixture.VBGMM
from whats new reported here
- Linear, kernelized and related models
  - residual_metric has been deprecated in :class:linear_model.RANSACRegressor. Use loss instead. By Manoj Kumar_.
  - Access to public attributes .X_ and .y_ has been deprecated in :class:isotonic.IsotonicRegression. By :user:Jonathan Arfa <jarfa>.
- Decomposition, manifold learning and clustering
  - The old :class:mixture.DPGMM is deprecated in favor of the new :class:mixture.BayesianGaussianMixture (with the parameter weight_concentration_prior_type='dirichlet_process'). The new class solves the computational problems of the old class and computes the Gaussian mixture with a Dirichlet process prior faster than before. :issue:7295 by :user:Wei Xue <xuewei4d> and :user:Thierry Guillemot <tguillemot>.
  - The old :class:mixture.VBGMM is deprecated in favor of the new :class:mixture.BayesianGaussianMixture (with the parameter weight_concentration_prior_type='dirichlet_distribution'). The new class solves the computational problems of the old class and computes the Variational Bayesian Gaussian mixture faster than before. :issue:6651 by :user:Wei Xue <xuewei4d> and :user:Thierry Guillemot <tguillemot>.
  - The old :class:mixture.GMM is deprecated in favor of the new :class:mixture.GaussianMixture. The new class computes the Gaussian mixture faster than before and some of computational problems have been solved. :issue:6666 by :user:Wei Xue <xuewei4d> and :user:Thierry Guillemot <tguillemot>.
- Model evaluation and meta-estimators
  - The :mod:sklearn.cross_validation, :mod:sklearn.grid_search and :mod:sklearn.learning_curve have been deprecated and the classes and functions have been reorganized into the :mod:sklearn.model_selection module. Ref :ref:model_selection_changes for more information. :issue:4294 by Raghav RV_.
  - The grid_scores_ attribute of :class:model_selection.GridSearchCV and :class:model_selection.RandomizedSearchCV is deprecated in favor of the attribute cv_results_. Ref :ref:model_selection_changes for more information. :issue:6697 by Raghav RV_.
  - The parameters n_iter or n_folds in old CV splitters are replaced by the new parameter n_splits since it can provide a consistent and unambiguous interface to represent the number of train-test splits. :issue:7187 by :user:YenChen Lin <yenchenlin>.
  - classes parameter was renamed to labels in :func:metrics.hamming_loss. :issue:7260 by :user:Sebastián Vanrell <srvanrell>.
  - The splitter classes LabelKFold, LabelShuffleSplit, LeaveOneLabelOut and LeavePLabelsOut are renamed to :class:model_selection.GroupKFold, :class:model_selection.GroupShuffleSplit, :class:model_selection.LeaveOneGroupOut and :class:model_selection.LeavePGroupsOut respectively. Also the parameter labels in the :func:split method of the newly renamed splitters :class:model_selection.LeaveOneGroupOut and :class:model_selection.LeavePGroupsOut is renamed to groups. Additionally in :class:model_selection.LeavePGroupsOut, the parameter n_labels is renamed to n_groups. :issue:6660 by Raghav RV_.
  - Error and loss names for scoring parameters are now prefixed by 'neg_', such as neg_mean_squared_error. The unprefixed versions are deprecated and will be removed in version 0.20. :issue:7261 by :user:Tim Head <betatim>.

files with remaining deprecated lines

amueller · 2017-08-16T21:50:33Z

(and yes, I'm just trying to hack my lines added / lines deleted ratio on github, you got me ;)

fixup! remove RandomizedPCA from docs references etc

Remove mixture/gmm

sklearn-lgtm · 2018-05-26T16:43:32Z

This pull request fixes 2 alerts when merging f114920 into 20cb37e - view on lgtm.com

fixed alerts:

1 for Non-callable called
1 for Non-iterable used in for loop

Comment posted by lgtm.com

sklearn-lgtm · 2018-06-04T18:04:58Z

This pull request fixes 2 alerts when merging fab56a3 into f049ec7 - view on LGTM.com

fixed alerts:

1 for Non-callable called
1 for Non-iterable used in for loop

Comment posted by LGTM.com

jnothman · 2018-06-05T07:50:35Z

CI failures

glemaitre · 2018-06-14T08:59:57Z

@amueller Do you mind if I am solving the conflicts and make the CI happy?

TomDLT · 2018-06-14T15:56:22Z

benchmarks/bench_plot_incremental_pca.py

-        results_dict = {k: benchmark(est, data) for k, est in [('pca', pca),
-                                                               ('rpca', rpca)]}
+        rpca = PCA(n_components=n_components, svd_solver='randomized', random_state=1999)
+        results_dict = {k: benchmark(est, data) for k, est in [('pca', pca)]}


This example does not work.
We should either keep rcpa here or remove it everywhere.

@TomDLT Is the suggestion of @jnothman a few lines above to use PCA(svd_solver='randomized') what you mean with keeping rpca ?

Yes, or we can just drop it.
My comment was meant to mention that the example is currently broken.

jorisvandenbossche

Did a quick skip through the diff as well, and given the earlier positive reviews, I would suggest to update this with master and merge it.
Then we can further investigate/clean-up remaining deprecation warnings on master.

jorisvandenbossche · 2018-06-22T08:11:59Z

sklearn/tree/export.py

        return '"tree.dot"'


 SENTINEL = Sentinel()


this sentinel class can be removed now

jorisvandenbossche · 2018-06-22T08:13:46Z

benchmarks/bench_plot_incremental_pca.py

-        results_dict = {k: benchmark(est, data) for k, est in [('pca', pca),
-                                                               ('rpca', rpca)]}
+        rpca = PCA(n_components=n_components, svd_solver='randomized', random_state=1999)
+        results_dict = {k: benchmark(est, data) for k, est in [('pca', pca)]}


@TomDLT Is the suggestion of @jnothman a few lines above to use PCA(svd_solver='randomized') what you mean with keeping rpca ?

jnothman · 2018-06-24T11:03:26Z

There is currently a test failure (as well as flake8 failure and merge conflicts):

=================================== FAILURES ===================================
 test_non_meta_estimators[GaussianProcessRegressor-GaussianProcessRegressor-check_estimators_unfitted] 
name = 'GaussianProcessRegressor'
Estimator = <class 'sklearn.gaussian_process.gpr.GaussianProcessRegressor'>
check = <function check_estimators_unfitted at 0x7f000a3d11b8>
    @pytest.mark.parametrize(
            "name, Estimator, check",
            _generate_checks_per_estimator(_yield_all_checks,
                                           _tested_non_meta_estimators()),
            ids=_rename_partial
    )
    def test_non_meta_estimators(name, Estimator, check):
        # Common tests for non-meta estimators
        estimator = Estimator()
        set_checking_parameters(estimator)
>       check(name, estimator)
Estimator  = <class 'sklearn.gaussian_process.gpr.GaussianProcessRegressor'>
check      = <function check_estimators_unfitted at 0x7f000a3d11b8>
estimator  = GaussianProcessRegressor(alpha=1e-10, copy_X_train=True, kernel=None,
        ..., normalize_y=False,
             optimizer='fmin_l_bfgs_b', random_state=None)
name       = 'GaussianProcessRegressor'
/home/travis/build/scikit-learn/scikit-learn/sklearn/tests/test_common.py:96: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
/home/travis/build/scikit-learn/scikit-learn/sklearn/utils/testing.py:328: in wrapper
    return fn(*args, **kwargs)
/home/travis/build/scikit-learn/scikit-learn/sklearn/utils/estimator_checks.py:1510: in check_estimators_unfitted
    est.predict, X)
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
exceptions = (<type 'exceptions.AttributeError'>, <type 'exceptions.ValueError'>)
message = 'fit'
function = <bound method GaussianProcessRegressor.predict of GaussianProcessRegressor(alp... normalize_y=False,
             optimizer='fmin_l_bfgs_b', random_state=None)>
args = (array([[-0.45364538, -0.47282444, -1.20608008, ..., -0.75500806,
         0.25... 0.45314754, -1.1924583 , ..., -1.72110924,
         0.27199858, -1.38161571]]),)
kwargs = {}, names = 'AttributeError or ValueError'
    def assert_raise_message(exceptions, message, function, *args, **kwargs):
        """Helper function to test the message raised in an exception.
    
        Given an exception, a callable to raise the exception, and
        a message string, tests that the correct exception is raised and
        that the message is a substring of the error thrown. Used to test
        that the specific message thrown during an exception is correct.
    
        Parameters
        ----------
        exceptions : exception or tuple of exception
            An Exception object.
    
        message : str
            The error message or a substring of the error message.
    
        function : callable
            Callable object to raise error.
    
        *args : the positional arguments to `function`.
    
        **kwargs : the keyword arguments to `function`.
        """
        try:
            function(*args, **kwargs)
        except exceptions as e:
            error_message = str(e)
            if message not in error_message:
                raise AssertionError("Error message does not include the expected"
                                     " string: %r. Observed error message: %r" %
                                     (message, error_message))
        else:
            # concatenate exception names
            if isinstance(exceptions, tuple):
                names = " or ".join(e.__name__ for e in exceptions)
            else:
                names = exceptions.__name__
    
            raise AssertionError("%s not raised by %s" %
>                                (names, function.__name__))
E           AssertionError: AttributeError or ValueError not raised by predict
args       = (array([[-0.45364538, -0.47282444, -1.20608008, ..., -0.75500806,
         0.25... 0.45314754, -1.1924583 , ..., -1.72110924,
         0.27199858, -1.38161571]]),)
exceptions = (<type 'exceptions.AttributeError'>, <type 'exceptions.ValueError'>)
function   = <bound method GaussianProcessRegressor.predict of GaussianProcessRegressor(alp... normalize_y=False,
             optimizer='fmin_l_bfgs_b', random_state=None)>
kwargs     = {}
message    = 'fit'
names      = 'AttributeError or ValueError'
/home/travis/build/scikit-learn/scikit-learn/sklearn/utils/testing.py:404: AssertionError

Any idea why this failure is occurring?

jnothman · 2018-06-24T11:22:51Z

I say we merge on green.

sklearn-lgtm · 2018-06-24T12:03:52Z

This pull request fixes 3 alerts when merging ee5710d into 62301aa - view on LGTM.com

fixed alerts:

1 for Non-callable called
1 for Unused import
1 for Non-iterable used in for loop

Comment posted by LGTM.com

jnothman · 2018-06-24T13:06:37Z

Thanks @amueller

jnothman · 2018-06-24T13:06:56Z

And thanks @massich!!

jnothman · 2018-06-24T13:08:01Z

Cue lots of complaints about cross_validation and grid_search disappearing... :)

jnothman · 2018-06-24T13:11:30Z

We just lost 10,200 lines of code :D

amueller · 2018-06-27T16:14:53Z

thanks for the fixes @jnothman! Sorry I was absent, I'm so glad this is in!

jnothman · 2018-06-27T23:26:57Z

no worries. I thought it was a good idea to merge *before* release :D

amueller · 2018-06-28T14:55:52Z

What's your preferred timeline now?

jnothman · 2018-06-28T21:18:34Z

My preferred timeline? I'm feeling very full up of work things at the moment and can't see myself being able to do anything focused towards release... but mostly there are a handful of things that we should still be trying to squeeze into release (deprecations, bug fixes, maybe a MissingIndicator, etc.), several of which are awaiting second review. The key features are in.

jnothman · 2018-06-28T21:19:48Z

I would also personally like to see some of #9599 merged to help libraries extend the search approach in BaseSearchCV...

amueller · 2018-06-29T20:35:27Z

I really would like the tags but not sure it's worth delaying the release...
Can you / have you flagged stuff for release?

glemaitre · 2018-06-29T20:38:55Z

We did flag the issues/PRs for the release with the 0.20 milestone.
Of course, we might miss some and some others could be to challenging.

qinhanmin2014 mentioned this pull request Nov 10, 2017

[MRG] MNT Remove code deprecated in 0.18 #10094

Closed

45 tasks

Joan Massich and others added 14 commits November 10, 2017 11:22

Remove deprecated (0.18) cross_validation.py in favor of model_selection

eb4d179

Fix imports (from corss_validation module to model_selection module)

6dfe9aa

Remove tests checking old implementation

af42424

Remove grid_search and learning_curve also deprecated

2362011

Remove gaussian_process

776bba1

remove code to be removed in 0.19

59e3f7d

remove ransac's residual_metric

2ec39c0

remove RandomizedPCA (also from docs references etc)

c444763

fixup! remove RandomizedPCA from docs references etc

remove references to old GP, GMM and sparse_center_data

a2e40d7

Remove mixture/gmm

more cleanup of deprecated scorers

7d4b2c1

More in scoring

2ffa7bd

Remove hamming_loss deprecated parameter classes

0bf4146

splitter classes (issue:6660) Fix minor stuff

b36341e

Fix doctest expected output

4b7aa69

amueller force-pushed the 0_20_deprecations branch from c9776c7 to 4b7aa69 Compare May 22, 2018 17:49

amueller added 4 commits May 22, 2018 13:54

merge

67d711f

unused imports

195fcf3

add vscode to gitignore

daa5e4b

delete files again after botched merge.

b72c9b9

amueller mentioned this pull request May 24, 2018

Common test to check that __init__ doesn't mess with parameters #11117

Closed

amueller added 7 commits May 24, 2018 12:27

import fix

f30720a

Merge branch 'master' into 0_20_deprecations

f114920

pep8

048a5ca

pep8

ddd45bd

delete old GMM

49b1498

remove deprecated scorers

3144d76

attributes X_, y_ in isotonic

ec66b5d

amueller added 2 commits June 4, 2018 13:21

Merge branch 'master' into 0_20_deprecations

5b8746b

keep randomized PCA in incremental benchmark.

fab56a3

glemaitre added this to the 0.20 milestone Jun 8, 2018

ogrisel mentioned this pull request Jun 14, 2018

Review DeprecationWarnings and FutureWarnings in raise tests for the 0.20 release #11252

Closed

TomDLT reviewed Jun 14, 2018

View reviewed changes

rth added the Blocker label Jun 14, 2018

jorisvandenbossche reviewed Jun 22, 2018

View reviewed changes

jnothman added 3 commits June 24, 2018 21:07

Revert change to estimator_checks about GaussianProcessRegressor

5ca5cfc

Merge branch 'master' into HEAD

4bb61c9

Clean up revert of removing rpca from benchmark

bdf90c4

Fix exception of GPR

ee5710d

jnothman merged commit eec7649 into scikit-learn:master Jun 24, 2018

amueller deleted the 0_20_deprecations branch June 27, 2018 16:14

Uh oh!

[MRG] 0.20 deprecations #9570

[MRG] 0.20 deprecations #9570

Uh oh!

Conversation

amueller commented Aug 16, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

files with remaining deprecated lines

Uh oh!

amueller commented Aug 16, 2017

Uh oh!

sklearn-lgtm commented May 26, 2018

Uh oh!

sklearn-lgtm commented Jun 4, 2018

Uh oh!

jnothman commented Jun 5, 2018

Uh oh!

glemaitre commented Jun 14, 2018

Uh oh!

TomDLT Jun 14, 2018

Choose a reason for hiding this comment

Uh oh!

jorisvandenbossche Jun 22, 2018

Choose a reason for hiding this comment

Uh oh!

TomDLT Jun 22, 2018

Choose a reason for hiding this comment

Uh oh!

jorisvandenbossche left a comment

Choose a reason for hiding this comment

Uh oh!

jorisvandenbossche Jun 22, 2018

Choose a reason for hiding this comment

Uh oh!

jorisvandenbossche Jun 22, 2018

Choose a reason for hiding this comment

Uh oh!

jnothman commented Jun 24, 2018

Uh oh!

jnothman commented Jun 24, 2018

Uh oh!

sklearn-lgtm commented Jun 24, 2018

Uh oh!

jnothman commented Jun 24, 2018

Uh oh!

jnothman commented Jun 24, 2018

Uh oh!

jnothman commented Jun 24, 2018

Uh oh!

jnothman commented Jun 24, 2018

Uh oh!

amueller commented Jun 27, 2018

Uh oh!

jnothman commented Jun 27, 2018 via email

Uh oh!

amueller commented Jun 28, 2018

Uh oh!

jnothman commented Jun 28, 2018

Uh oh!

jnothman commented Jun 28, 2018

Uh oh!

amueller commented Jun 29, 2018

Uh oh!

glemaitre commented Jun 29, 2018

Uh oh!

Uh oh!

amueller commented Aug 16, 2017 •

edited

Loading