Subject data has been preprocessed using the cpac pipeline and globally filtered.

pheno_train, pheno_test -- Contain SUB_ID and labels ( DX_GROUP) for training  + lots and lots of metadata -- most important columns have self-explanatory names.

abide.hdf5: contains functional connectivity data for all subjects in efficient binary format

train, test : contain original functional connectivity data in 1D format
sample_submission: example submission format

Good luck!