摘要：Medoid-based method is an alternative technique to centroid-based method for partitional clustering algorithms. This method has been incorporated in a recently introduced clustering algorithm for categorical data, called k-Approximate Modal Haplotype (k-AMH) algorithm. This study reports the performance evaluation between the medoid-based method represented by the k-AMH algorithm and the centroid-based method represented by the extended k-Mode algorithm, the k-Population algorithm and the new Fuzzy k-Mode algorithm in clustering common categorical data. Nine common categorical data sets were used in the experiments to compare the performance of both methods using clustering accuracy scores. In overall results, the medoid-based method of k-AMH algorithm produced significant results for all data sets. The method showed its advantage of obtaining the highest clustering accuracy of 0.94 when clustering large number of clusters. This result indicated that the medoid-based method has a significant contribution for clustering categorical data, particularly for clustering large number of clusters.
