League Championship Algorithm for clustering

In the last decade the nature inspired algorithms have gained a lot of popularity in solving complex optimization problems. Partitional clustering deals with the optimization of data points from the cluster centroids to classify a dataset into several groups (clusters). In this paper we introduce clustering as an optimization problem and solve it with a recently developed natural meta-heuristic League Championship Algorithm (LCA). The performance of the proposed algorithm is compared with benchmark K-means, Particle Swarm Optimization, CLONAL algorithm of Artificial Immune System. The superior performance of the proposed algorithm is demonstrated over its counterpart models in terms of percentage of accuracy and computational time over fifty independent runs.