ON DISAGGREGATE TRAVEL MODELS, AND SOME THOUGHTS ON MODELS IN GENERAL

The report discusses the general form of disaggregate travel models, which are an important recent development in transport modelling. The traditional aggregate model uses data averaged over geographical zones, and unless great care is used in its application, grossly misleading relationships can be deduced between the explanatory variables and the travel decisions they predict. The solution is to use A disaggregate model which uses data on individual journeys in its formulation. A commonly-used disaggregate model is the multinomial logistic model, but in this model it is necessary to enumerate the travel options and this can cause problems when two options are similar but not identical. A better type of model is one based on minimum-generalised costs and employing a cost distribution. The traveller is then assumed to travel by the option with the least generalised cost, and it is not necessary to include the number of alternative options in the model. The report concludes that the best type of travel model is one based on minimum generalised cost and disaggregated data. (TRRL)