WBBA-KM: A Hybrid Weight-Based Bat Algorithm with K-Means Algorithm For Cluster Analysis

Abstract

Data clustering is an unsupervised classification method used to classify unlabeled objects into clusters. The clustering is performed by partitioning clustering, hierarchical clustering, fuzzy clustering, and density-based clustering methods. However, the center of the clusters is updated according to local searches with these traditional methods, and finding the best clusters center affects the clustering performance positively. In this study, a variant bat algorithm called weight-based bat algorithm (WBBA) is proposed and the proposed WBBA hybridized with the k-means clustering method (WBBA-KM) to determine the optimal centers of the clusters. The performance of the proposed WBBA-KM has been evaluated by using six different benchmark datasets from the UCI repository and the obtained results are compared with FCM, IFCM, KFCM, KIFCM, PSO-IFCM, GA-IFCM, ABC-IFCM, PSO-KIFCM, GA-KIFCM, ABC-KIFCM, and BA-KM clustering methods in the literature. According to the experimental results, the proposed WBBA-KM clustering method performed better performance from all other clustering methods in 4 of 6 benchmark datasets and achieved better performance from the BA-KM clustering method in all benchmark datasets.