Data Mining and Knowledge Engineering

Abstract

Clustering analysis is a challenging task and there area number of issues associated with it, e.g. accuracy, quality, efficiency, finding cluster of different shape, size and density finding clusters which are sensitive to noise and outliers. K-means clustering algorithms are widely used for many practical applications. Original k-mean algorithms select initial centroids randomly that affect the accuracy and quality of the resulting clusters and sometimes it generates poor and empty clusters which are meaningless. Our approach for the K-mean algorithm eliminates the deficiency of exiting K-mean algorithm and Improve accuracy and generate high quality cluster by reducing mean square error.

Keywords

Full Text