[2M4-OS-37a] OS-37

[2M4-OS-37a-01] Decomposition and estimation of latent structure underlying people counting data in indoor space

〇Kei Aizawa¹ (1. TAKENAKA CORPORATION)

Keywords:Spatial statistical model, Bayesian estimation

For strategic location planning, it is important to understand the latent structure of attracting customers. Although various area marketing tools for analyzing the latent structure in outdoor space have appeared, there are few tools for that in indoor space. In recent years, it has become easier to measure people counting data in indoor space, so that the people counting data can be used for analyses in indoor space. In this study, I developed a statistical model which explains the people counting data from spatial and non-spatial elements and sensor arrangements. Then, I estimated the model with real-world data. The results show that the latent structure was decomposed into spatial effects of sub-spaces in the indoor space, and non-spatial effects such as the day of the week and the shop composition. Investigating the spatial effects in detail, I found areas which are expected to attract many customers.