Understanding Customer Behavior in Shopping Mall from Indoor Tracking Data

The prosperity of various indoor positioning technologies makes possible the large collection of tracking data in indoor spaces. Much of the focus has been on several fundamental problems such as indoor localization, indoor space modeling, indoor data cleansing, indexing and querying. However, this paper attempts to analyze customer behavior from a unique indoor tracking data, which will promote the convergence between various applications and the underlying data. In particular, we introduce a real-life indoor tracking data collected at an electrical mall in China. Then, we cluster users into several groups and summarize the most characteristic behaviors of each cluster. Last but not least, we analyze customer's individual behaviors through two aspects: 1) the regression model is used to reveal hot regions where customers are likely to stay in long time; and 2) a transition matrix combined with the connectivity is presented to demonstrate hot paths.