Practical algorithms for clustering and modeling large data sets : analysis and improvements