Skip to Main Content
Clustering the process of grouping homogeneous objects is an important data mining process. Few algorithms exist to cluster categorical data. K-modes is the scalable and efficient algorithm to cluster the categorical data. In this paper we propose a new distance measure for K-modes based on the cardinality of domain of attribute. The proposed method is experimented with data sets obtained from UCI data repository. Results prove that the proposed measure generates better clusters than the K-modes algorithm.