Disaggregate modelling, which attempts to explain the choices made by individual travellers, has now become established as an appropriate procedure for the analysis of many issues in transport planning that require travel demand forecasts. Similarly, the technique of sample enumeration, the `natural' forecasting procedure for use with disaggregate models, is now often used for short-term forecasting in the original study area. However, the application of sample enumeration for longer-term forecasting or for forecasting in other areas is less well established. A method that has proved successful for transferring models to different location or time in a number of studies is prototypical sample enumeration. This method generates a sample of individuals that is as far as possible consistent with the information that is available about the larger area or time, often using aggregate data, but is also as far as possible consistent with the detailed interactions present in the original sample used to estimate the model. The paper explains the procedures of prototypical sampling, incorporating a number of variants to accommodate different degrees of availability of data. The various optimisation processes that can be used are specified and the use of each in differing circumstances is justified. The potential for error and error reduction are examined. Examples of the procedure and model applications are given from a number of studies in various parts of Europe.
Samenvatting