A hybrid GA-GSA noval algorithm for data clustering

Clustering is widely used in data mining and machine learning to group the objects of a given dataset based on similarities between data without using any previous knowledge for the purpose to extract the crucial information. In this paper a noval clustering algorithm is developed by integrating two well known optimization technique “genetic algorithm (GA) and gravitational search algorithm (GSA)”. Hybrid GA-GSA performance is tested against K-means, GA and GSA clustering algorithm for seven popular dataset iris, wine, breast cancer, vowel, crude oil, glass and cmc taken from uci machine learning repository over three criterion i.e. sum of intra cluster distance, error rate and CPU execution time. Experimental results tabulated in Table II–V show hybrid GA-GSA algorithm for clustering is a better technique against K-means, GA and GSA.