Time-series modelling for forecasting vehicular traffic flow in Dublin

Abstract

The traffic flow at an arterial intersection in a congested urban transportation network in the city of Dublin is modelled in this paper. Three different time-series models, viz. random walk model, Holt-Winters’ exponential smoothing technique and seasonal ARIMA model are used for modeling of traffic flow in Dublin. Simulation and short-term forecasting of univariate traffic flow data are done using these models. The data used for modeling are obtained from loop-detectors at a certain junction in the city center of Dublin. Seasonal ARIMA and Holt-Winters’ exponential smoothing technique give highly competitive forecasts and match considerably well with the observed traffic flow data during rush hours.