A k Nearest Neighbor based Local Linear Wavelet Neural Network Model for On-line Short-term Traffic Volume Prediction

摘要: This paper presents a forecasting method called k nearest neighbor based local linear wavelet neural network (kNN-LLWNN) for the on-line, short-term prediction of five-minute traffic volumes at westbound Interstate 64 in Hampton Road Virginia. The is on combining (k-NN), with (LLWNN). idea to apply k-NN form training dataset LLWNN instead taking whole historical training. proposed model compared k-NN, and support vector regression (SVR) separately from accuracy running time two aspects. Experiments are conducted decide most appropriate parameters four models verification dataset. For test dataset, study's findings appear confirm hypothesis that, kNN-LLWNN performs comparable SVR, its much lower than SVR because introduction k-NN.