Forecast based on different data types: A before and after study (Stated preference Mode choice)

Abstract

In addition to adequate data, the formulation of transport forecasts relies upon a knowledge of the relationships between the demand for transport and those factors which influence it. These relationships are described with mathematical functions and the their model parameters. The parameters derived from revealed preference (RP) data are often subject to too many impon-derables and are thus of only limited value in many cases. The main cause of this uncertainty is data which is either insufficiently detailed or unsuitable for estimating transport demand functions. For this reason, an earlier SVI study concerning the sensitivity of passenger transport to sup-ply- side and price changes recommended that research should be conducted in parallel with major transport infrastructure investments. In this way this recommended project would be able to check and validate the quality of the findings from alternative or supplementary stated preference (SP) (Vrtic et al., 2000), as both forecast and actual changes in demand would be known to it. The launch of intercity tilting-trains (known as ICNs) in 2001 and further supply-side im-provements to road and rail transport supply offered an opportunity to conduct ex ante/ex post surveys in order to verify the forecasting approaches in a defined period. This mix of qualita-tive and quantitative changes is a particular challenge for forecasting and the attendant data collection process. Nonetheless, it is a challenge that must be overcome again and again in day-to-day practice. As the supply-side changes are generally small, it was expected that changes on the demand side will be concentrated at the level of mode and route choice. The principal aim of this research remit was to verify and identify the limits and possibilities of the three data sources for forecasting by means of an ex ante/ex post analysis. However, this study also offered an opportunity to analyse other aspects of importance to transport fore-casting. Here, in addition to the study methodology, the quality and accessibility of the avail-able bodies of data proved to be crucial factors in modelling transport movements and events. In a first stage the study estimated a detailed public transport route choice model and cali-brated national network models for both road and rail demand. This is an essential preliminary stage to the calculation of modal shifts in demand for transport and the subsequent review of the different forecasts. In the case of mode choice changes, the three most common ap-proaches to forecasting were to be tested: • Direct elasticity, known from previous studies • RP models, i.e. model parameters based on RP data • SP models, i.e. model parameters based on SP data The primary benefits of this study can be summarized as follows: - It sets out the opportunities and limits, as well as the advantages and disadvan-tages, of the three data sources for forecasting under review. - This the first study to provide models of route and mode choice which have been estimated from SP data. The model parameters estimated using this data provide the basis for the practical application of mode and route choice models following supply-side transport changes. - The estimated model parameters, current figures and the relative valuations dem-onstrate the importance of individual variables to mode and route choice. They were estimated for each trip purpose. - The study showed that in this case the forecasts derived from the SP-data were more consistent and more precise than either the estimates from the direct elastic-ities or the RP data. - Verifying transport forecasts shows how and where further improvements can or must still be made with regard to both data bases and methodology. - The study describes the possibilities and methodical foundation for common SP/RP estimates of model parameters.