Use sklearn train_test_split in place of StratifiedShuffleSplit

[rhiever]

The latest version of sklearn's train_test_split automatically stratifies the data by class(es) and shuffles the data prior to the split. Thus, we don't need to use the long-form version with StratifiedShuffleSplit.

Let's replace all of the StratifiedShuffleSplit calls with train_test_split.

[pronojitsaha]

Ok..will look into this.

[rhiever]

Ping. How's this coming along? Should I mark it as "being worked on"?

[pronojitsaha]

Yes, I did give it a try couple of days back, but getting some spurious results. Checking on it. yes you can mark it.

[pronojitsaha]

Have sent across the PR. Thanks.