On selecting the initial cluster centers in the K-means algorithm

K-means clustering algorithm which is a process of separating n number of points into K clusters according to the predefined value of K is one of the clustering analysis algorithms. This algorithm has many applications in analysis of clustering. There are many factors that affect performance of the K-means clustering algorithm to better cluster. One of these is selecting initial cluster centers. In this study, two methods have been proposed for selecting the initial cluster centers. The proposed methods have been tested on data sets taken from UCI database and compared with the method proposed by Erisoglu etc and K-means algorithm which generates initial centers randomly. The comparison results show that the K-means algorithm which uses the proposed methods converges to better clustering results. © 2017 IEEE.