Customer Segmentation in Electronics Retail Using Self-Organizing Maps

Abstract

Customer segmentation in marketing research has the aim of gathering more knowledge of the customer base to provide appropriate services to individual groups according to their needs and desires. This paper presents an approach to customer segmentation using self-organizing maps (SOM). The SOM algorithm projects high-dimensional input space onto nodes on a two-dimensional grid that can be utilized to visually explore the data and examine the relationships between data features. Hierarchical clustering is applied to the nodes on the SOM grid to give a suitable number of segments which represent the whole customer base. The segmentation model gives a visual output that is easily interpretable for decision-makers. An empirical study using historical transaction data from an electronics retailer is conducted and the segmentation model is further supported by applying the method to the customer base of a grocery retailer. The SOM algorithm and the consequential cluster analysis was implemented using R.