A shopping-path length estimation using Markov-chain-based shopper dynamics model

We propose a framework of estimating shopping-path length, in which the floor is represented by the graph G(V,E) with a vertex set V and an arc set E and the shopping-path length is measured by the number of zones (vertices) shoppers visit. We used the Markov-chain to model the dynamics of distribution of shoppers on the vertecies in the graph. We derive the (discrete) probability distribution of shopping path length using the transition matrix in the Markov-chain, and derive the expected path length. We proposed the index called the improvement importance index to quantify how local changes in the transition probability affect the entire shopping path length. We have tested our framework to the test data from an industrial application and the estimated path-length is compared to the actual one. We have a result that the error of estimation is 0.2%.