Short-Term Traffic Flow Forecasting Using ARIMA-SVM Algorithm and R

Traffic flow forecasting is an important issue for route guidance and traffic flow control in intelligent transportation system. Real-time and reliable traffic flow information is directly related to the effect of traffic management and control. In this paper, considering the nonlinear characteristics of traffic flow caused by the complexity of time-varying and changing uncertainties in the actual traffic system, we use support vector machine (SVM) model to predict short-term traffic flow. Then combining the autocorrelation of traffic flow, an improved prediction model combining ARIMA and SVM is proposed. Finally, the model is validated by using the measured traffic data and R language. Simulation results show that the improved ARIMA-SVM model can enhance the accuracy of traffic flow forecasting greatly and describe the law of traffic flow exactly, and it has a wide application prospect in short-term traffic flow and other nonlinear prediction problems.