A Novel Approach for Initial Centroid Computation in Clustering
Clustering is one of main aspects in wireless sensor networks and K-means algorithm is most efficient way to cluster analysis. K-means algorithm, forms the group of clusters which based on the random selection of initial centroids. If initial centroids are selected based up on distribution of clusters in that particular area, then better set of clusters can be obtained. This paper proposes a method based on the rectangular area division algorithm to find the better initial centroid which forms more accurate cluster with considerably reducedcomputing time. The analysis and experimental output shows that the proposed rectangular area division method should increase the accuracy of clusters and reduce the computational time of the K-means algorithm.
Author's Name: Allan J Wilson, Dr.A.S. Radhamani and Dr.P. Kannan
Volume: Volume 15
Issues: Issue 3
Keywords: Cluster Analysis, K-Means Algorithm, Data Sets, Cluster Centroids, Rectangular Area Division, Computational Time.