train_test_split documentation: add example for stratify param

[kirisakow]

Reference Issues/PRs

None

What does this implement/fix? Explain your changes.

Provided an understandable example to the train_test_split function documentation, namely concerning the stratify parameter. Inspired by this reply on StackOverflow.

[NicolasHug]

Thanks for the PR @kirisakow , we usually don't include these details in the docstring, but rather in the user guide.

In this case I think we can just add a "For more details on stratification, see the <User Guide -- Link here> " and link to https://scikit-learn.org/dev/modules/cross_validation.html#cross-validation-iterators-with-stratification-based-on-class-labels

We may also add a note that train_test_split behaves exactly like SSS in this subsection: https://scikit-learn.org/dev/modules/cross_validation.html#stratified-shuffle-split

[kirisakow]

Thank you for explanation and counter-proposal @NicolasHug. I was actually afraid my suggestion would meet a snobbish reaction and would be dismissed

[cmarmo]

Hi @kirisakow, thanks for your pull request.
There are linting issues in your code. In the lint details you can see the message

sklearn/model_selection/_split.py:2121:80: E501 line too long (81 > 79 characters)

[NicolasHug]

Thanks @kirisakow , minor suggestion on top of @cmarmo 's comment. Otherwise LGTM

[kirisakow]

@cmarmo, @NicolasHug, done!

Please, accept my apologies for my clumsiness.

[NicolasHug]

Thanks @kirisakow !

[kirisakow]

Thank YOU, both of you, for your kind support!