RFM Based Market Segmentation Approach Using Advanced K-means and Agglomerative Clustering: A Comparative Study

Abstract

Customer segmentation based on RFM (Recency, Frequency & Monetary) variables is very popular nowadays to partition the customers pursuant to their characteristics and behaviour. Several clustering algorithms have been implemented to segment customers into groups to achieve better clustering results. In this paper a comparative analysis among agglomerative, k-means and advanced version of k-means are carried out for RFM based market segmentation approach. The experimental outputs show that the agglomerative clustering needs a long processing time for large dataset in comparison with k-means and advanced version of k-means clustering. However one crucial advantage of agglomerative clustering is, it does not require to appoint the total number of clusters initially. The segmentation outcomes also exhibit that the advanced version of k-means can effectively reduce the total running time by 27.8% and 97.8% compared to standard k-means and agglomerative clustering respectively and so effectively increases the speed of clustering. It can also produce better clustering results than standard k-means in respect of intra cluster distance and inter cluster distance. Our proposed method of clustering based customer segmentation uses advanced version of k- means which reduces the overall time complexity and performs quite well for large dataset.