Hybrid Bisect K-Means Clustering Algorithm

Abstract

In this paper, we present a hybrid clustering algorithm that combines divisive and agglomerative hierarchical clustering algorithm. Our method uses bisect K-means for divisive clustering algorithm and Unweighted Pair Group Method with Arithmetic Mean (UPGMA) for agglomerative clustering algorithm. First, we cluster the document collection using bisect K-means clustering algorithm with the value K', which is greater than the total number of clusters, K. Second, we calculate the centroids of K' clusters obtained from the previous step. Then we apply the UPGMA agglomerative hierarchical algorithm on these centroids for the given value, K. After the UPGMA finds K clusters in these K' centroids, if two centroids ended up in the same cluster, then all of their documents will belong to the same cluster. We compared the goodness of clusters generated by bisect K-means and the proposed hybrid algorithms, measured on various cluster evaluation metrics. Our experimental results shows that the proposed method outperforms the standard bisect K-means algorithm.