Customer mobility and congestion reduction in supermarkets

The analysis and characterization of human mobility using population-level mobility models is important for numerous applications, ranging from commuting flows to modeling trade flows. However, almost all of these applications have focused on large spatial scales, varying from intra-city level to country level. In this work, we study population-level mobility models in supermarkets, and we thereby examine these models on a much smaller spatial scale. We use a novel data set to infer origin–destination trips in supermarkets and apply variants of the gravity model and the intervening-opportunities model to fit this trip data and predict trip distributions from unseen data. We find that a production-constrained gravity model and an extended radiation model (a variant of the intervening-opportunities model) can successfully predict 65–70% of the trips inside supermarkets.
We then use the models to find how to reduce congestion in supermarkets. Given a store layout, we apply the extended radiation model to identify store layouts with less congestion. We use two popular measures for congestion on networks: the maximum number of visits to a node and the total mean queue size. We are able to find store layouts with less congestion than the original layout with either measure using a simulated-annealing algorithm. These new layouts suggest dispersing zones with high number of sales across the store (see Fig. 1). Our work therefore gives insight both into how customers move in supermarkets and into how retailers can arrange stores to reduce congestion.