Clustering Bank's Customers Using Artificial Neural Networks

Abstract

Customer's Clustering is an instrument for considering the needs which were not allowed to be expressed due to mass marketing. The primary goal of market segmentation is to find and retain those customers we want to serve. In this paper, we present the experimental results of clustering bank's customers using artificial neural networks (ANN) compared with traditional statistical methods. To cluster the customers, 7 key distinctive characteristics of 600 customers of a bank were extracted. Customers’ clustering was performed using ANN and a powerful statistical method: Ward Method. The results were compared using discriminant analysis, MAPE and RMSE. The comparisions indicate the superiority of ANN output over WARD. Clustering by ANN indicates the strength and innovation of this study. Furthermore, focusing on importance of solving the death neuron problem in artificial neural networks by BIAS term is a contribution of this paper.

Keywords

• Customer's Clustering
• Dead unit
• Discriminant analysis
• Haming
• Ward Method