How to formulate an accident prediction model for urban intersections.

It is generally accepted that accident rates tend to be higher at intersections than on through sections of a road. This is particularly frequent in urban area where roads are characterized by intersections in close succession; moreover, the safe and effective operations of the urban road system can be significantly affected by safety conditions at intersections. In this paper models and methods designed to understand and to predict the accident process at urban intersections are reviewed. In particular this study intends to show why the statistical modelling approach is useful for accident analysis and how it can be applied to provide some general advices for conducting safety evaluations with accident data. An exploratory example describes how to formulate an accident model for urban intersections by the analysis of significant explanatory variables affecting accident phenomenon and by modelling of accident and traffic data for urban intersections. At last, in view of the recent great interest in safety problems of urban roads with particular regard to intersections, the research intends to summarize the main features of the accident intersection models and their share in developing quantitative safety effectiveness measures for installation design improvements.