The influence of sales areas and bargain sales on customer behavior in a grocery store

Developments in radio frequency identification (RFID) technology have resulted in the availability of data on customers' movement paths in various stores. In this paper, we propose a customer behavior model in a grocery store by using RFID and point-of-sales data. This model is based on a nonhomogeneous hidden Markov model with covariates and estimates "Stop" and "Pass by" behaviors. The model introduces sales areas and the number of bargain products as covariates and quantifies the effect of these covariates on each behavior. Thus, we can diagnose sales areas and decide the optimal quantity of bargain products. Further, we can rearrange sales areas and reinforce weak sales areas according to the diagnosis results. In addition, information on the optimal quantity of bargain products allows implementation of an effective bargain sales strategy.