Finding Customer Loyalty Based on Weighted RFMD Clustering Model

Abstract

Customer loyalty is an important aspect of customer relationship management. An organization needs to analyze customer loyalty in order to understand and retain their valuable customers. Earlier methods to calculate customer loyalty based on RFM model resulted in poor accuracy. Therefore, a new procedure has been proposed in this work to find customer loyalty by including an important attribute called 'duration'. Work presented in this paper is focused at retail business. Retail Data Transaction dataset has been taken from Kaggle repository. The RFMD attributes of buyers have been extracted from this dataset. To find the most loyal customers, the dataset is subjected to clustering based on weighted attributes. Relative weights are assigned to the RFMD attributes, calculated using the AHP method. A RFMD comparison matrix has been formed from evaluators response and relative weights of attributes are calculated by using Eigen Vector method. Clustering is applied on RFMD attributes of buyers using k-means algorithm and then Customer Value of each cluster is calculated. The cluster having the highest Customer Value represents the highest degree of loyalty. Results are evaluated by calculating classification accuracy using Decision tree and KNN. The proposed method has been found more accurate in determining loyalty of customers compared to earlier methods.