Predictive Reliability of Airport Ground Access Mode Choice Models

Abstract

Airport ground access mode choice models provide an essential analytical capability to support airport ground transportation planning studies as well as studies addressing the allocation of air travel demand among multiple airports in a region. A growing number of such models have been described in the literature: the models vary widely in functional form, explanatory variables, and estimated parameter values. However, almost no attention has been given to how reliably these models predict air passenger mode choice for time periods different from those for which they have been estimated or when available ground access services change. Both considerations are critical to the application of such models for forecasting. In attempting to rectify this situation, the predictive reliability of typical airport ground access models is addressed from two perspectives: (a) the variability of airport ground access mode use over time and implications of this variation for the application of ground access mode choice models and (b) how well models estimated on data collected at one time predict air traveler ground access mode use at other times or in different situations. The requirements for an empirical analysis that would apply an existing model to predict mode use in a later year or for a different airport are discussed, and the predicted mode use is compared with the actual mode use observed in air passenger surveys to assess the predictive reliability of the model.