Support for abstract cv method for StackingCVClassifier

[sque]

Description

Change the API of StackingCVClassifier in order to accept any kind of cross-validation technique.

Related issues or pull requests

Related #202

Pull Request requirements

• Added appropriate unit test functions in the ./mlxtend/*/tests directories
• Ran nosetests ./mlxtend -sv and make sure that all unit tests pass
• Checked the test coverage by running nosetests ./mlxtend --with-coverage
• Checked for style issues by running flake8 ./mlxtend
• Added a note about the modification or contribution to the ./docs/sources/CHANGELOG.md file
• Modify documentation in the appropriate location under mlxtend/docs/sources/ (optional)
• Checked that the Travis-CI build passed at https://travis-ci.org/rasbt/mlxtend

[coveralls]

Coverage increased (+0.02%) to 93.68% when pulling 7bc008f on sque:abstract_cv_method_for_stacking into e688f7d on rasbt:master.

[rasbt]

Looks nice, but could you mention that the n_folds argument was changed to cv? Maybe sth. like

• The n_folds parameter of the StackingCVClassifier was changed to cv and can now accept any kind of cross validation technique that is available from scikit-learn. For example, StackingCVClassifier(..., cv=StratifiedKFold(n_splits=3)) or StackingCVClassifier(..., cv=GroupKFold(n_splits=3)) (via Konstantinos Paliouras).

[sque]

I agree it looks better like this. Thanks

[rasbt]

Thanks for the PR, the code looks fine, and that's a pretty elegant solution. Regarding the examples in the documentation (at https://github.com/rasbt/mlxtend/blob/master/docs/sources/user_guide/classifier/StackingCVClassifier.ipynb). Are you planning on changing the previous once to make sure they work with the API change? Also, it would prob. be a good idea to add a new example for GroupKFold or so.

[coveralls]

Coverage increased (+0.02%) to 93.68% when pulling 0cbd4d1 on sque:abstract_cv_method_for_stacking into e688f7d on rasbt:master.

[sque]

I updated the CHANGELOG as you instructed but I would like to avoid updating the documentation because I am not familiar with the process. In case I am commiting more changes in the future, I will try to learn the process :)

[rasbt]

Thanks! For the documentation, you would just need to update the Jupyter Notebook text. However, I just see that the n_folds parameter was nowhere used explicitly in the examples, so nothing would have to be changed here :). The only think that would change is the API documentation at the bottom, but this will be done automatically when I push the documentation to the website during the next version release.

The general way would be to go into the docs subdirectory and execute the make_api.py script from there:

cd docs
python make_api.py

This would build a local API documentation via markdown, which is then simply read into the Jupyter notebook -- later, the Jupyter notebook gets converted into Markdown via

python ipynb2markdown.py --ipynb sources/user_guide/classifier/StackingCVClassifier.ipynb

But don't worry about it, I will run ipynb2markdown.py --all before the next version release.

Overall, it looks good now, I can merge it now unless you have additional suggestions

[sque]

Thank you for the info and I will keep them in mind for the next time :) I do not intend to make any more changes you can proceed with the merging

[rasbt]

Nice, I will merge it then! Thanks a lot for the contribution, I greatly appreciate it!