Study of extension of k-means paradigm to clustering mixed numerical and categorical data