A Chaotic Charged System Search Approach for Data Clustering

Abstract

Data clustering is a key technique in the field of data mining, pattern recognition, bioinformatics and machine learning which concerns the organization and unexplored relationship between the huge amounts of data. It can analyse the data without knowing the size and distribution. Thus, this paper presents a new approach based on the charged system search (CSS) with chaotic map for partition clustering. The aim of this method is to achieve the global optimum solution by minimizing an objective function. The chaotic charge system search algorithm (CCSAA) utilizes the concept of CSS algorithm and chaos theory to obtain the desired results. The quality of the proposed algorithm is evaluated on seven datasets and then compared with other well-known algorithms in data mining domain. From the simulations results, it is observed that the proposed algorithm delivers more efficient and effective results than the other methods being compared.