An accessibility-based model is now being developed for the Knoxville (TN) Transportation Planning Organization, based on the success of innovations in and research related to the development of a model for the Northwest Arkansas Regional Planning Commission. These accessibility-based models represent a new framework and model design which is distinct from both traditional four-step trip-based models and activity-based simulation models.The approach is trip-based and produces deterministic forecasts, like a traditional model. However, it uses a disaggregate synthetic population, represents time choices in a pseudo-continuous and behaviorally sensitive manner and incorporates some simultaneity of considerations, even in choices regarding different trips. Statistically, the approach avoids both the aggregation bias endemic to traditional models as well as the random simulation errors introduced by activity-based models which complicate forecasting and alternatives analyses. Behaviorally, accessibility-based models are able to reflect trip-chaining and other complex behaviors like activity-based models but with far fewer component models and without explicitly representing activities or tours. Hence, they incur computational and development costs more similar to four-step models. Although they cannot offer the full behavioral richness of activity-based models, they present a major improvement over traditional four-step models at only a fraction of the cost of an activity-based application, making them an appealing choice for many agencies with a desire for greater behavioral realism but a limited modeling budget.
Samenvatting