Automatic Clustering for Customer Segmentation by Adaptive Differential Evolution Algorithm

Abstract

One of the main issues in cluster analysis is to determine the correct number of clusters in the real-world applications. This study suggests an adaptive Differential Evolution (DE) algorithm to perform the automatic clustering and determine the number of clusters automatically. The proposed approach is denoted as Adaptive Differential Evolution (ADE) utilizes adaptive approaches to fine-tune the standard DE algorithms key parameters and uses new mutation approaches to keep a balance between the exploration and exploitation in the algorithm. The proposed algorithm is used for several benchmark datasets and applied to customer segmentation as a case study. After performing the customer segmentation, different clusters are obtained which will determined different groups of customers. In order to test the efficiency of the proposed approach, the Wilcoxon Rank sum test has been used. After conducting the experiments, the proposed algorithm revealed a superior efficiency in comparison with the employed algorithms.