New datasets

[denkle]

The first attempt to start adding datasets from a collection used within “Do we Need Hundreds of Classifiers to Solve Real World Classification Problems?”
The file for the first dataset is one of the most important ones because other files from the collection will pretty much follow what is specified in this file.

[mikeheddes]

Thanks for submitting this PR! It looks great, I think having all these datasets as part of the library is a great addition and from here it should not be too hard to add more of them. Great work!

[mikeheddes]

This is looking really good, exactly what I had in mind. I just added some minor code organization and refactoring comments. Which is mostly about trying to isolate the common behavior.

Also, can you remove the .DS_Store file from the PR? and make sure to add Adult also to the documentation.

[denkle]

This file should not have been deleted - I did that accidentally while trying to revert its inclusion into the PR.
Not sure at the moment how to revert this deletion.

[mikeheddes]

I am resolving some minor outstanding issues and will push my changes soon. Small question, is the number of folds always 4 or is it dataset dependent?

[denkle]

I am resolving some minor outstanding issues and will push my changes soon. Small question, is the number of folds always 4 or is it dataset dependent?

Yes, for datasets in the collection the number of folds is always 4.

[mikeheddes]

@denkle could you review my refactoring of the _load_data methods? I want to make sure I didn't break it. Otherwise I think it's good to go

[denkle]

@denkle could you review my refactoring of the _load_data methods? I want to make sure I didn't break it. Otherwise I think it's good to go

@mikeheddes, great revision of the code! The logic is more streamlined in multiple places! I do not see any problems with _load_data methods so assume it is good to go

[denkle]

Good to go, I believe