Abstract
It is important to analyse pedestrian traffic line with regard to planning and evaluating walking spaces. In this paper, we focus on angular data and discuss a method to analyse pedestrian traffic line. We regard traffic line as data consisted of pedestrian’s position and angle: direction of movement at each time. Basic analysis shows that many pedestrians draw a similar trajectory around a ticket gate at a station; direction of movement is mostly determined by a relative position to the ticket gate. This fact means that a position of pedestrian can explain an angle at that position. Therefore we try to model this relation, using a method developed in directional statistics. In case of a simple dataset we successfully estimated a regression model of angle, which satisfied the property of angular data such as cyclicity. At last we considered some applications of the estimated model, taking an anomaly detection and an angle-choice model for example.
