Efficient Identification of Initial Clusters Centers for Partitioning Clustering Methods
Ram Pal Singh, Dharmveer Singh Rajpoot · 2019
K-Means is an important and widely used partitioning method of clustering, but it has some drawback like random initialization of center, which may affect the final cluster center. To reach the final cluster center it takes more time as usual. There are many algorithms for the initial center problem and some of them are very useful like k-means++ and canopy, Hartigan-Wong, MacQueen, etc. These algorithms are used in Rstudio and Weka which are very strong tools of machine learning, especially in clustering. In this paper, we propose an algorithm for initializing the centers of k-means and the proposed method is compared with other six well know existing methods and three experiments are done on three data sets and results are compared using clustering parameters. It is found that the proposed method works better as compared to other existing methods.