Predicting People Flow for Supporting Facility Management
Keisuke Tsunoda, Takahiro Hata, Kazuaki Obana · 2020
This paper proposes a novel method to predict people flow in a facility such as a shopping mall for supporting facility management including air-conditioning, cleaning, maintenance, and support for tenant shops or restaurants. Existing studies have attempted to estimate or predict people flow several minutes or hours in the future or only occupancy level. However, facility managers need to know people flow including the number of visitors, their direction of movement, and their average speed up until several days before because they must decide details of air-conditioning, cleaning, and maintenance several days in advance. To support their daily facility management, we propose a novel method to predict the number of moving people and their average speed in each defined direction in a facility the next day or later on the basis of characteristics of past people flow on weekends and weekdays. We evaluated the effectiveness of our proposal using measured data in the shopping mall.