In the present report we aim at providing a critical view to some existing time-series models for road safety accident prediction in order to allow for a better picture of the wide scope of the problem and also to be able to criticize approaches. A special focus is given to models that allow for disaggregate analysis, i.e. analysis that focus on different subgroups of the road safety population. Apart from simple time series models and discrete valued models, the currently fashionable and powerful state space models are investigated in more detail as they prove to be a promising technique for further analysis. Some issues related to data availability are also discussed. Conclusions and general comments with respect to the selection of appropriate models to forecast accident rates for different groups in Flanders for the forthcoming years are given. (Author/publisher)
Samenvatting