Development of a Traffic Accident Prediction Model for Urban Signalized Intersections

It is commonly estimated that there is a much higher potential for accidents at a crossroads than along a single road due to its plethora of conflicting points. According to the 2006 figures by the National Police Agency, the number of traffic accidents at crossroads is greatly increasing compared to that along single roads. Among others, crossroads installed with traffic signals have more varied influential factors for traffic accidents and leave much more room for improvement than ones without traffic signals; thus, it is expected that a noticeable effect could be achieved in safety if proper counter-measures against the hazards at a crossroads were taken together with an estimate of causes for accidents This research managed to develop models for accident forecasts and accident intensity by applying data on accident history and site inspection of crossroads, targeting four selected downtown crossroads installed with traffic signals. The research was done by roughly dividing the process into four stages: first, analyze the accident model examined before; second, select variables affecting traffic accidents; third, develop a model for traffic accident forecasting by using a statistics-based methodology; and fourth, carry out the verification process of the models.