Accident prediction models for urban areas - a state-of-the-art

Abstract

This paper summarizes the results of the bibliographic study on accident prediction models applied to urban areas. Several types of models where analyzed, namely for accidents involving pedestrians or cyclists, collisions between motorized vehicles, total accidents, non injury accidents, accidents with fatalities, injury accidents, night time accidents and accidents involving only vehicles. Models with different levels of disagregation were also studied: aggregated models describe general safety trends (national or regional); disaggregated models represent specific changes in the transportation system. This study is part of the “IRUMS – Safer Roads in Urban Areas” project, carried out at the National Laboratory of Civil Engineering and at the Department of Engineering of the University of Coimbra, financed by the Foundation for Science and Technology. This project intends to develop methods for safety management of urban road networks. Procedures for the estimation of expected accident frequencies, identification of sites with a promise and selection of efficient corrective measures are being developed as well. The case study is being applied in Lisbon.