Performance Analysis of Data Clustering Algorithms using Various Effectiveness Measures

Krishnamoorthi Murugasamy and natarajan Mathaiyan

Department of Computer Science and Engineering, Bannari Amman Institute of Technology, India            

Abstract: Data clustering is a method to group the data records that are similar to each other. In recent days, researcher show significant attention towards the use of swarm based optimization algorithms to improve the performance of clustering process. This Performance analysis concentrates on the effectiveness of five different algorithms with respect to various distances metrics to find the effective algorithm among them. The algorithms used for comparison are K-means algorithm, Artificial Bee Colony (ABC) algorithm, Fuzzy C-Means (FCM) incorporated ABC (ABFCM) algorithm, K-means incorporated Artificial Bee Colony (ABK) algorithm and Bacterial Foraging Optimization Algorithm. Among those algorithms, ABFCM and ABK algorithms are enhanced ABC algorithm in which the FCM and K-means operator are incorporated in the scout phase of the traditional ABC algorithm respectively. In this paper, the performance of these algorithms are compared in terms of various distances metrics like dice coefficient, jaccard coefficient, beta index and distance index by varying the cluster sizes and number of iteration. Finally, from the experimental results it proves that the proposed algorithms ABFCM and ABK outperforms better when compared with the existing algorithms.

Keywords: Data clustering, k-means algorithm, FCM, ABC, distances metrics.

Received November 5, 2012; accepted February 24, 2013