Remove duplicated columns (resulting from a join)

Added train, test, validation split for both normal dataset as well as the new rank for a subset of the dataset