Short-time traffic flow forecasting based on projection pursuit autoregression

Accurate short-time traffic flow forecasting is one of the important issues for Intelligent Transportation Systems research, especially for the Advanced Traffic Management Systems and Advanced Traveler Information Systems research. With the shortening of the forecasting term, the uncertainty of traffic flow becomes more and more seriously, so that the forecasting effect of general approaches is decreasing. For an example, the algorithm based on nonparametric regression is a real-time non-parametric forecasting algorithm with the characteristic of high transplantation and accuracy, which plays an important role in traffic flow forecasting, yet there is the problem of dimension curse as the dimensions of the sample data increase. For the purpose of solving the question of short-time traffic flow forecasting, a short-time traffic flow forecasting model based on projection pursuit autoregression technique is established in this paper. The problem of dimension curse and non-normality among high-dimensions data are solved. This algorithm satisfies the need of real-time traffic flow forecasting completely through the field data test.