Are Zero-suppressed Binary Decision Diagrams Good for Mining Frequent Patterns in High Dimensional Datasets?