Interval prediction for traffic time series using local linear predictor

This paper addresses the issue of the interval forecasting (constructing prediction intervals for future observations) of the traffic data time series using one of local polynomial nonparametric models - the local linear predictor. Two methods are proposed and compared. One is based on the theoretical formulation of the asymptotic prediction intervals and another is an empirical procedure using bootstrap, both for the local linear predictor. Finally, a case study using real-world traffic data is presented for both approaches, along with the results compared with each other. The results coincide with expectations and have validated the proposed methods.