Research issues on K-means Algorithm: An Experimental Trial Using Matlab

Clustering problems arise in many different applications: machine learning data mining and knowledge discovery, data compression and vector quantization, pattern recognition and pattern classification. It is considered that the k-means algorithm is the best-known squared errorbased clustering algorithm, is very simple and can be easily implemented in solving many practical problems. This paper presents the results of an analysis of the representative works related to the reseach lines of k-means algorithm devoted to overcome its shortcomings. To establish a framework for a proposed improvement to the standard k-means algorithm, the results obtanained from experiments of the kmeans in the Matlab package and databases of the UCI repository are