Title

Planning Constrained Destination Choice in the ADAPTS Activity-Based Model

Abstract

This seminar will present the Agent-based Dynamic Activity Planning and Travel Scheduling (ADAPTS) model framework recently which has recently been developed. This model focuses on simulating the dynamics of the underlying planning processes behind daily activity scheduling. As the model focuses on the dynamics of activity scheduling formation, new scheduling constraints are introduced to the model, referred to as planning constraints, that depend on the order in which activities enter the daily plan. This concept is demonstrated by focusing on a destination choice model estimated from recent Chicago survey data coupled with a model of the timing of activity planning decisions. The travel survey data is used to estimate both a standard multinomial logit destination choice model, and a conditional choice model under certain assumptions about the activity planning process, where the choice set is constrained by what has already been planned in the schedule. The performance of each model is evaluated and the impact of using the planning-constrained model in place of the standard model on the accuracy of the results is evaluated. The use of a model where the destination choices are conditioned on what was previously planned in each activity-travel schedule improves the accuracy of the model results. This is true in terms of correctly predicted location choices and overall trip length distributions, with much more realistic distributions observed when destination decisions are constrained.