Analysis of hourly traffic volumes on expressway considering cyclic variation and autoregression

Abstract

For the high quality infrastructure in the future, it is important to construct and manage road network , based on the traffic characteristics. It is necessary to understand the characteristics of traffic for that purpose. In this decade, it became easier to handle the consecutive data of traffic, so that we analyze the traffic volume of expressway using state-space model. We assume three factors that affect the state of traffic volume. One is the seasonal trend, the others are the "day of the week" effect and the "time of day" effect. Autoregression is also included to consider the overflow effect by the road capacity, which restricts the traffic volume at congestion time. The algorithm to estimate the state of traffic is Kalman filter that allows us to estimate the state by sequential analysis, which can consider new data addition. As the result of the state variables, we saw the characteristics of the traffic and easily found the difference of the purpose of traffic on each IC.