Comparative Study of Initial Centroid based K-Mode Algorithm

Abstract

 Clustering is one of the techniques of the data mining, which defines classes and put the objects into one
group having similar properties and objects having dissimilar properties into another group. An extension of the KMeans Algorithm, K-Mode Algorithm, is partitioning based clustering algorithm but it does not guarantee for the optimal
solution. In this paper, there is the comparative analysis of Ini_Distance and Ini_Entropy Algorithm with Cao’s methods,
WK-Mode with Chan’s Algorithm, Harmonic K-Mode with K-Mode and EC K-Mode Algorithm on real datasets. These
algorithms are based on the selection of initial centroids in which the clustering accuracy is improved. The algorithms
discussed in this study can be improved further by other better optimization techniques through research made in this
field.