Traffic Accident Macro Forecast Based on ARIMAX Model

Abstract

To overcome the deficiency of traditional traffic accident estimation models, this paper introduced a new way. It gave two categories on traffic accident affected factors and selected the main ones, using stepwise regression model. ARIMAX model, a dynamic regression one, was used to forecast traffic accident volumes. The former job ensures the precision of estimation, while the latter one owns both regression and ETS models’ merits. The example of traffic accident data from 1983 to 2005 in China was taken to validate the feasibility of the model. It established the relationship between dead people due to traffic accident and total population, mileage of road rank, the number of passenger transport, the population of drivers, the average GDP and automobile numbers, and gave an estimation adopting ARIMAX model. The result shows that the error is small and it has a much better foreground on traffic accident macro forecast.