Application of Support Vector Regression and Particle Swarm Optimization in Traffic Accident Forecasting

Abstract

Traffic accident forecasting is important for altering and planning of road. Recently time series analysis is an important direction in traffic accident forecasting. Support vector regression (SVR) a kind of SVM used in regression and has better nonlinear forecasting performance than BP neural network. In the paper, the combination method based on particle swarm optimization and support vector regression (PSO-SVR) is adopted in traffic accident forecasting, and particle swarm optimization (PSO) is introduced to choose the parameters of SVR and improve the forecasting performance of SVR. The experimental results indicate that the proposed PSO-SVR method has better results than BP neural network in the traffic accident forecasting.