Jaydee

I believe I found a workaround for this. Considering these problems tend to happen to highly imbalanced datasets, I would suggest using StratifiedShuffleSplit, and alter the train_size or test_size ratio...