{"title":"Enhancing K-Means Algorithm with Initial Cluster Centers Derived from Data Partitioning along the Data Axis with the Highest Variance","authors":"S. Deelers, S. Auwatanamongkol","country":null,"institution":"","volume":11,"journal":"International Journal of Physical and Mathematical Sciences","pagesStart":518,"pagesEnd":524,"ISSN":"1307-6892","URL":"https:\/\/publications.waset.org\/pdf\/4854","abstract":"In this paper, we propose an algorithm to compute\r\ninitial cluster centers for K-means clustering. Data in a cell is\r\npartitioned using a cutting plane that divides cell in two smaller cells.\r\nThe plane is perpendicular to the data axis with the highest variance\r\nand is designed to reduce the sum squared errors of the two cells as\r\nmuch as possible, while at the same time keep the two cells far apart\r\nas possible. Cells are partitioned one at a time until the number of\r\ncells equals to the predefined number of clusters, K. The centers of\r\nthe K cells become the initial cluster centers for K-means. The\r\nexperimental results suggest that the proposed algorithm is effective,\r\nconverge to better clustering results than those of the random\r\ninitialization method. The research also indicated the proposed\r\nalgorithm would greatly improve the likelihood of every cluster\r\ncontaining some data in it.","references":null,"publisher":"World Academy of Science, Engineering and Technology","index":"Open Science Index 11, 2007"}