Segmenting Consumer Location-Product Preferences for Assortment Localization

Abstract

When managing multiple stores in the same marketplace, retailers need to select store locations and localize product assortments to reflect the demand of each community it serves. This paper develops a dynamic system, called the dual Poisson-Gamma Dynamic Systems (dPGDS), for panel data on product assortments and individual consumers’ purchases across store/vending locations. The dPGDS can help retailers automatically profile different consumer segments driven by store visiting preferences, measure the relationships across store locations, and estimate the product preferences for each consumer segment simultaneously. The dPGDS relies on a Bayesian nonparametric prior and can be efficiently trained for large-scale transactional data using our proposed MCMC inference algorithm. We apply the dPGDS in the retail vending market in major train stations in Japan. We demonstrate the face validity of the dPGDS and its value for retailers to improve vending location decisions as well as product assortment at the location level.