Data Clustering Based on Data Transformation and Hybrid Step Size-Based Cuckoo Search

Abstract

Data clustering is a prominent analytic method which discovers the clusters in a dataset based on some similarity measures. Since, clustering is a NP-hard problem, thus it is difficult to find optimal cluster for large and high dimensional dataset using traditional methods. Therefore, this paper introduces a data clustering method based on data transformation and hybrid step size based cuckoo search method (HSCS). Data transformation method uses PCA and ICA for transformation of data which is further used by HSCS method for clustering. The performance of the proposed method has been tested on the six benchmark datasets taken from UCI repository and compared with cuckoo search, particle swarm optimization, differential evolution, and improved cuckoo search.