Comparative Analysis of Hybrid K-Mean Algorithms on Data Clustering

Abstract

Data clustering is a process of organizing data into certain groups such that the objects in the one cluster are highly similar but dissimilar to the data objects in other clusters. K-means algorithm is one of the popular algorithms used for clustering but k-means algorithm have limitations like it is sensitive to noise ,outliers and also it does not provides global optimum results. To overcome its limitations various hybrid k-means optimization algorithms are presented till now. In hybrid k-means algorithms the optimization techniques are combined with k-means algorithm to get global optimum results. The paper analyses various hybrid k-means algorithms i.e. Firefly, Bat with k-means algorithm, ABCGA etc. The Comparative analysis is performed using different data sets obtained from UCI machine learning repository. The performance of these hybrid k-mean algorithms is compared on the basis of output parameters like CPU time, purity etc. The result of Comparison shows that which k-means hybrid algorithm is better in obtaining cluster with less CPU time and also with high accuracy.