Advantages and disadvantages of different crash modeling techniques.

Modeling of traffic crashes is a complex undertaking. This short paper, from the Traffic Records Forum, held in Buffalo, NY in 2005, reports on a study of the advantages and disadvantages of different crash modeling techniques. The authors review the methods that have been used, including regression analysis, artificial neural networks, and pattern recognition methods that include nearest neighbor rule and Bayesian belief network technique. The authors conclude that each method has its advantages and disadvantages and that the suitability of the method depends upon the desired output and the model inputs. The output of the model could include type of the crash (fatal, injury, and property damage only, number of crashes, crash rate, segment/intersection severity rating); the inputs may include different environmental, traffic, and roadway variables.