Data Clustering via Principal Direction Gap Partitioning