Highway Traffic Volume Forecasting Based on Seasonal ARIMA Model

Abstract

In order to improve the accuracy of seasonal highway traffic volume forecasting, a general expression of seasonal ARIMA model with periodicity was presented based on the normal ARIMA model, and then the procedures of modeling and forecasting via seasonal ARIMA model were provided. In the feasibility-study experiment, the seasonal length was calculated by Fourier period analysis method, the test of the stationarity of the time series, and identifying, establishing, choosing and forecasting of the model were done by Eviews software. Compared with three normal seasonal forecasting models (grouped regression model, variable seasonal index forecasting model and seasonal regression model), the seasonal ARIMA model can obtain the highest accuracy in forecasting. The research result is significant to forecast highway traffic volume more accurately.