A statistical consistency method for evaluating the output from traffic simulation and forecasting models

A post evaluation of a forecasting model involves comparing these outputs with real data. This is only possible as far as the forecasting horizon. This paper is to be able to ascribe a level of confidence to predicted values. The approach we have used is based on work conducted by meteorologists. The study described in this paper uses the concept of "ensemble forecasting systems" efs to conduct a pre-evaluation of traffic forecasts made one year in advance by the French National Road Information Centre, " Bison Fute" system. We describe also how to apply this evaluation approach to the output from traffic simulation models.