The Application of Nonparametric Regressive Algorithm for Short-Term Traffic Flow Forecast

Abstract

Short-term traffic flow forecast is an important topic in the research field of intelligent transportation systems. The article analyses the preliminary results in the short-term traffic flow forecast, takes full advantage of the characteristics of K-neatest neighbor (KNN) classifiers, and builds a model based on nonparametric regressive algorithm.The historical and metrical data is classified by KNN,and the state vector is constructed by utilizing the output of KNN classifier. Traffic flow forecasting for the next period is entirely based on the state vectors.The experimental results show that the model was verified more accurate.