Note train_auto is the training data, test_auto is the test dataset, mean_auto is a sample submission.
Data fields are index and p_target for mean_auto and for all submissions.  All files must be csv.