A novel spatial clustering analysis method using bat algorithm

In data mining, clustering can be regarded as the process of finding some optimal centers. Bat algorithm is a powerful method for solving many multi-objective optimization problems. It has fast convergence and global optimization ability of the search features. In this paper, we present a bat clustering algorithm (BCA) based on similarity, and improve the corresponding fitness function. Contrary to Particle Swarm Optimization (PSO) clustering algorithm using Euclidean distance, BCA performs a more correct search in the entire solution space. Spatial data sources are from IRIS dataset and wine dataset. These data are clustered by clustering analysis based on BCA-similarity algorithm. Experimental results show that BCA- similarity clustering method can achieve fast and accurate spatial data clustering.