Data mining method for incident duration prediction

Abstract

Traffic incident management and information dissemination strategies will benefit from the prediction of incident duration in real time. This study investigates the development of an incident duration prediction model based on a detailed historical incident database. A data mining technique, namely the Bayesian Network was applied to develop the prediction models. The analysis results suggest that the Bayesian Network model is advantageous in terms of accurate prediction and the convenience of application. © 2011 Springer-Verlag.