A Human-Flow Analysis Based on PCA:A Case Study on Population Data Near Railway

The damage caused by natural disasters and accidents is increasing every year. To reduce such damage from spreading, it is important to detect an accident promptly. However, current sensing systems are difficult to use because they have narrow coverage and are specialized in few detectable accidents. In this paper, we propose a method to detect disasters and accidents by calculating the degree of an anomaly in human flow by treating a common human flow as a single large sensor. Human flow can be assumed to have typical patterns in people’s daily life, such as going to work and leaving work. Developing an anomaly detection method of human flow can lead to the discovery of any hidden causes such as accidents and disasters. In this paper, we study a method that aims to detect anomalies in human flow, considering the operational status of railways as an example. We confirm that our method can detect the actual suspension of operations.