Development of accident prediction models for pedestrian crossings

Abstract

In large Polish cities like Warsaw, pedestrians constitute almost 60% of road fatalities. Although traffic safety situation in general is improving, the numbers of pedestrians hit when crossing a road have not significantly decreased over the last six years. A negative binomial model was estimated for predicting accidents at unsignalised pedestrian crossings based on accident data from 52 crossings in Warsaw. A total of 58 pedestrian accidents were recorded at these crossings during the last seven years. The model shows that the number of accidents is less-than-proportional to both pedestrian and motorised traffic daily volumes. Other risk factors affecting pedestrian safety are: higher proportion of heavy vehicles and location in a mixed land use area. The model can be used with the Empirical Bayes method for an unbiased identification of high risk locations.