Accidents are caused by faulty decisions of traffic participants, often partially influenced by the traffic situation itself. There are specific situations in which numerous accidents occur; these accidents can be studied and the circumstances improved accordingly. However, the majority of accidents occur outside these so-called black spots. Other traffic situations are characterized by the fact that practically no accidents occur in a period of a year. Nevertheless, because there is a countless number of such situations, the total number of accidents is high. It is virtually impossible to make a study of accidents in such isolated situations. Regression model techniques do provide a feasible approach to this type of traffic situation. A model that describes the effects of the traffic and road characteristics can be used to determine the accident rate. Instead of waiting until accidents occur which can then be analysed, a forecast is made of the accidents that can be anticipated. Systematic and area-wide reconstruction regulations can then be enforced to reduce the probability of such accidents occurring. This contribution deals with a regression model technique that is suitable for use in such cases. This technique, which has been successfully applied in studies on interurban bicycle and car traffic, is based on the weighted multiproportional Poisson model. For the covering abstract of the conference see IRRD abstract no 264967.
Abstract