Analysis of Supermarket Customer Behavior by Using RFM and Huffs Gravity Model

Abstract

This paper discusses how to evaluate supermarket competition through distance data between addresses of customers of target supermarket chain and all supermarkets in a city and ID-POS data of target supermarket chain. We also detect the degree of influence on specific customers who are affected by the competition of other nearby supermarkets based on logistic regression analysis. In this paper, we consider how to transform the distance data positive correlation to customer affected degree to make the analysis correctly since the closer the customers' address to other supermarkets, the more likely they are influenced by these supermarkets, resulting in a decreased purchasing in target supermarket. In addition, we consider how to solve multicollinearity problems for training data. In numerical experiments, we comprehensively evaluate the influence of the other nearby supermarkets on customers purchasing of target supermarket from the viewpoint of partial regression coefficients and accuracy.