K-harmonic means Data Clustering using Combination of Particle Swarm Optimization and Tabu Search

Clustering is one of the widely used techniques for data analysis. Also it is a tool to discover structures from inside of data without any previous knowledge. K-harmonic means (KHM) is a center-based clustering algorithm which solves sensitivity to initialization of the centers which is the main drawback of K-means (KM) algorithm, but, both KM and KHM converge to local optimal. In this paper, a hybrid data clustering algorithm based on KHM is proposed called PSOTSKHM, using Particle Swarm Optimization (PSO) algorithm as a stochastic global optimization technique and Tabu Search (TS) algorithm as a local search method. This algorithm makes full use of the advantages of three algorithms. The proposed algorithm has been compared with KHM, PSOKHM and IGSAKHM algorithms on four real datasets and the obtained results show the superiority of suggested algorithm in most cases.