Performance Comparison of Social Spider Optimization for Data Clustering with Other Clustering Methods

In this paper, we evaluate a nature inspired algorithm namely Social Spider Optimization (SSO) for data clustering that performs local and global searches in a balanced way. We compare CPU elapsed time of the proposed algorithm SSO with K-means and Particle Swarm Optimization (PSO). Though the convergence speed of K-Means clustering is very high when compared with other clustering algorithms, it has the weakness of getting stuck in the local optima in the solution space. Though SSO solves this problem to an extent, it has to be combined with K-means in order to speed up the convergence. Hence, we combine SSO with K-means that produces good results in local searches. We compare CPU elapsed time of proposed hybrid algorithm SSO + K-means (SSOKC) with K-means + Bat algorithm (KBat), K-means + Firefly algorithm (KFA) and K-means + Flower pollination algorithm (KPA). The study is carried out on real life datasets taken from UCI machine learning repository.