Planning-Constrained Destination Choice in Activity-Based Model

This paper describes a destination choice model estimated for Chicago, Illinois, with the use of a recent survey and a previously developed model of activity planning decision timing. The travel survey data are used to estimate a standard multinomial destination choice model and a model in which the choice set is constrained by what has already been planned in the schedule. The performance of each model is evaluated, as is the effect that using the planning-constrained model in place of the standard model has on the accuracy of the results. Using a model in which the destination choices are conditioned on what has been previously planned improves the accuracy. This improvement is true for correctly predicted location choices, especially for overall trip length distributions in which more realistic distributions are observed when decisions are constrained.