[MRG] Fix FutureWarnings in logistic regression examples by ogrisel · Pull Request #12114 · scikit-learn/scikit-learn

ogrisel · 2018-09-19T16:38:25Z

Since #11905 we get warnings in when using LogisticRegression without setting an explicit solver or an explicit multi_class option.

I use LBFGS and multinomial as default and SAGA when it makes sense.

…t models

…probas.py

ogrisel · 2018-09-19T16:42:03Z

BTW, early stopping on the validation set would be nice ;)

ogrisel · 2018-09-19T16:44:46Z

examples/compose/plot_digits_pipe.py

+# Define a pipeline to search for the best combination of PCA truncation
+# and classifier regularization.
+logistic = SGDClassifier(loss='log', penalty='l2', early_stopping=True,
+                         max_iter=10000, tol=1e-5, random_state=0)


I use SGD classifier here because it's much faster thanks to the early stopping and we are only interested in generalization (not optimization and coef values) in this specific example.

TomDLT · 2018-09-19T16:46:02Z

examples/compose/plot_digits_pipe.py

-plt.axis('tight')
-plt.xlabel('n_components')
-plt.ylabel('explained_variance_')
+fig, (ax0, ax1) = plt.subplots(nrows=2, sharex=True, figsize=(4, 5))


We can't read y labels in the plot.
We need a different figsize, or to use plt.tight_layout().

ogrisel · 2018-09-20T09:51:51Z

.circleci/config.yml

+      - SCIPY_VERSION: "0.16"
+      - MATPLOTLIB_VERSION: "1.4"
+      - SCIKIT_IMAGE_VERSION: "0.11"
+      - PANDAS_VERSION: "0.17.1"


I want to use DataFrame.nlargest to make it easy to plot the results of a grid search. This causes a general bump up in the dependency versions for the documentation build. But I think this is fine.

… example doc

ogrisel · 2018-09-20T10:03:53Z

Thanks for the feedback @TomDLT . Hopefully I fixed both the rendering issue you reported and the python 2 build.

amueller · 2018-09-24T17:22:56Z

thanks. Generally I'm against tight_layout now that there's an actual layouting engine.

…n#12114) * Be more specific about logistic regression solver in examples * Use early stopped SGD (faster) and plot cross-validated error for best models * Fix LR solver in /plot_voting_probas.pyexamples/ensemble/plot_voting_probas.py * Fix LR solver & scale data in plot_digits_classification_exercise.py * Use saga solver in plot_logistic_l1_l2_sparsity.py * Use LBFGS solver in plot_iris_logistic.py * Use LBFGS in plot_logistic.py * Use SAGA solver for Logistic Regression Path example * Use LBFGS solver in plot_classifier_chain_yeast.py * Use LBFGS solver in plot_rbm_logistic_classification.py * typo * typo * Bump up pandas dependency to 0.17.1 * Bump up examples minimal deps to match pandas 0.17.1 * Fix figure layout for plot_digits_pipe.py * Version numbers are not decimal numbers * Set multinomial, no scaling to keep example simple, fix formatting of example doc * Missing plt.tight_layout() in plot_voting_probas.py * Missing plt.tight_layout() in plot_logistic.py

ogrisel · 2018-09-25T07:50:22Z

Generally I'm against tight_layout now that there's an actual layouting engine.

I am not familiar with the new layout API. What would be the best way to ensure that the axis labels are not cropped?

* tag '0.20.0': (77 commits) ColumnTransformer generalization to work on empty lists (scikit-learn#12084) add sparse_threshold to make_columns_transformer (scikit-learn#12152) [MRG] Convert ColumnTransformer input list to numpy array (scikit-learn#12104) Change version to 0.20.0 BUG: check equality instead of identity in check_cv (scikit-learn#12155) [MRG] Fix FutureWarnings in logistic regression examples (scikit-learn#12114) [MRG] Update test_metaestimators to pass y parameter when calling score (scikit-learn#12089) DOC Removed duplicated doc in tree.rst (scikit-learn#11922) [MRG] DOC covariance doctest examples (scikit-learn#12124) typo and formatting fixes in 0.20 doc (scikit-learn#11963) DOC Replaced the deprecated early_stopping parameter with n_iter_no_change. (scikit-learn#12133) [MRG +1] ColumnTransformer: store evaluated function column specifier during fit (scikit-learn#12107) Fix typo (scikit-learn#12126) DOC Typo in OneHotEncoder DOC Update fit_transform docstring of OneHotEncoder (scikit-learn#12117) DOC Removing quotes from variant names. (scikit-learn#12113) DOC BaggingRegressor missing default value for oob_score in docstring (scikit-learn#12108) [MRG] MNT Re-enable PyPy CI (scikit-learn#12039) MNT Only checks warnings on latest depedendencies versions in CI (scikit-learn#12048) TST Ignore warnings in common test to avoid collection errors (scikit-learn#12093) ...

* releases: (77 commits) ColumnTransformer generalization to work on empty lists (scikit-learn#12084) add sparse_threshold to make_columns_transformer (scikit-learn#12152) [MRG] Convert ColumnTransformer input list to numpy array (scikit-learn#12104) Change version to 0.20.0 BUG: check equality instead of identity in check_cv (scikit-learn#12155) [MRG] Fix FutureWarnings in logistic regression examples (scikit-learn#12114) [MRG] Update test_metaestimators to pass y parameter when calling score (scikit-learn#12089) DOC Removed duplicated doc in tree.rst (scikit-learn#11922) [MRG] DOC covariance doctest examples (scikit-learn#12124) typo and formatting fixes in 0.20 doc (scikit-learn#11963) DOC Replaced the deprecated early_stopping parameter with n_iter_no_change. (scikit-learn#12133) [MRG +1] ColumnTransformer: store evaluated function column specifier during fit (scikit-learn#12107) Fix typo (scikit-learn#12126) DOC Typo in OneHotEncoder DOC Update fit_transform docstring of OneHotEncoder (scikit-learn#12117) DOC Removing quotes from variant names. (scikit-learn#12113) DOC BaggingRegressor missing default value for oob_score in docstring (scikit-learn#12108) [MRG] MNT Re-enable PyPy CI (scikit-learn#12039) MNT Only checks warnings on latest depedendencies versions in CI (scikit-learn#12048) TST Ignore warnings in common test to avoid collection errors (scikit-learn#12093) ...

* dfsg: (77 commits) ColumnTransformer generalization to work on empty lists (scikit-learn#12084) add sparse_threshold to make_columns_transformer (scikit-learn#12152) [MRG] Convert ColumnTransformer input list to numpy array (scikit-learn#12104) Change version to 0.20.0 BUG: check equality instead of identity in check_cv (scikit-learn#12155) [MRG] Fix FutureWarnings in logistic regression examples (scikit-learn#12114) [MRG] Update test_metaestimators to pass y parameter when calling score (scikit-learn#12089) DOC Removed duplicated doc in tree.rst (scikit-learn#11922) [MRG] DOC covariance doctest examples (scikit-learn#12124) typo and formatting fixes in 0.20 doc (scikit-learn#11963) DOC Replaced the deprecated early_stopping parameter with n_iter_no_change. (scikit-learn#12133) [MRG +1] ColumnTransformer: store evaluated function column specifier during fit (scikit-learn#12107) Fix typo (scikit-learn#12126) DOC Typo in OneHotEncoder DOC Update fit_transform docstring of OneHotEncoder (scikit-learn#12117) DOC Removing quotes from variant names. (scikit-learn#12113) DOC BaggingRegressor missing default value for oob_score in docstring (scikit-learn#12108) [MRG] MNT Re-enable PyPy CI (scikit-learn#12039) MNT Only checks warnings on latest depedendencies versions in CI (scikit-learn#12048) TST Ignore warnings in common test to avoid collection errors (scikit-learn#12093) ...

ogrisel added 10 commits September 19, 2018 12:09

Be more specific about logistic regression solver in examples

d189026

Use early stopped SGD (faster) and plot cross-validated error for bes…

4b7483f

…t models

Fix LR solver in /plot_voting_probas.pyexamples/ensemble/plot_voting_…

447d6f2

…probas.py

Fix LR solver & scale data in plot_digits_classification_exercise.py

c54c342

Use saga solver in plot_logistic_l1_l2_sparsity.py

c4bba06

Use LBFGS solver in plot_iris_logistic.py

dfb94e2

Use LBFGS in plot_logistic.py

c856d9e

Use SAGA solver for Logistic Regression Path example

03102e0

Use LBFGS solver in plot_classifier_chain_yeast.py

3509ac1

Use LBFGS solver in plot_rbm_logistic_classification.py

ec2918e

ogrisel added this to the 0.20 milestone Sep 19, 2018

typo

30b56e5

ogrisel commented Sep 19, 2018

View reviewed changes

ogrisel added 2 commits September 19, 2018 18:47

typo

e76d5fc

Bump up pandas dependency to 0.17.1

7658268

TomDLT approved these changes Sep 20, 2018

View reviewed changes

ogrisel added 3 commits September 20, 2018 11:31

Bump up examples minimal deps to match pandas 0.17.1

0801606

Fix figure layout for plot_digits_pipe.py

8134fa8

Version numbers are not decimal numbers

2539fc5

ogrisel commented Sep 20, 2018

View reviewed changes

Set multinomial, no scaling to keep example simple, fix formatting of…

c7a5c05

… example doc

ogrisel changed the title ~~Fix FutureWarnings in logistic regression examples~~ [MRG] Fix FutureWarnings in logistic regression examples Sep 20, 2018

ogrisel added 2 commits September 20, 2018 13:28

Missing plt.tight_layout() in plot_voting_probas.py

2bbc6cd

Missing plt.tight_layout() in plot_logistic.py

79a97f9

amueller merged commit fc7d6e6 into scikit-learn:master Sep 24, 2018

amueller mentioned this pull request Sep 24, 2018

Additions to 0.20 between RC and release #11948

Merged

ogrisel deleted the fix-logistic-regression-examples branch September 25, 2018 07:42

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[MRG] Fix FutureWarnings in logistic regression examples#12114

[MRG] Fix FutureWarnings in logistic regression examples#12114
amueller merged 19 commits intoscikit-learn:masterfrom
ogrisel:fix-logistic-regression-examples

ogrisel commented Sep 19, 2018

Uh oh!

ogrisel commented Sep 19, 2018

Uh oh!

ogrisel Sep 19, 2018

Uh oh!

TomDLT Sep 19, 2018

Uh oh!

ogrisel Sep 20, 2018

Uh oh!

ogrisel commented Sep 20, 2018

Uh oh!

amueller commented Sep 24, 2018

Uh oh!

ogrisel commented Sep 25, 2018

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Uh oh!

Conversation

ogrisel commented Sep 19, 2018

Uh oh!

ogrisel commented Sep 19, 2018

Uh oh!

ogrisel Sep 19, 2018

Choose a reason for hiding this comment

Uh oh!

TomDLT Sep 19, 2018

Choose a reason for hiding this comment

Uh oh!

ogrisel Sep 20, 2018

Choose a reason for hiding this comment

Uh oh!

ogrisel commented Sep 20, 2018

Uh oh!

amueller commented Sep 24, 2018

Uh oh!

ogrisel commented Sep 25, 2018

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants