Destination Choice Models

A destination choice model is an econometric model that can be used to understand and predict the choice of trip destination, or in other words, the choice of activity location at the destination end of a trip. They can be used alone or as part of a regional travel demand model system. For instance, the trip distribution model in a four-step Urban Transportation Modeling System (UTMS) could be represented by a destination choice model when the analysis is undertaken at the disaggregate level. (A UTMS could operate at the aggregate level of trips between zones or at the disaggregate level of trips made by individual travelers.) Destination choice models can also be used to predict the locations of activity stops in an activity-based transportation model and are referred to as location choice models in this context. The analysis of short-term mobility behavior, in the form of daily activity and travel patterns, is thus the primary purpose of destination choice models.

The choice of destination is typically a specific building (for example, work or restaurant location) or a set of buildings (for example, a shopping complex or area). However, for transportation planning purposes, destination choice models usually operate with zonal choice alternatives, which are spatial clusters of elemental alternatives. The study area is divided into zones, for instance, transportation analysis zones (TAZs) or census divisions (tracts, blockgroups, blocks), which are described by attributes such as area, population, accessibility, land use composition, distance from the individual's home or work locations.

The zonal approach is preferred because (1) the use of elemental destination alternatives would impose a much larger computational burden, (2) data is rarely if ever available at the elemental level, and (3) for transportation planning, the desired objective is the prediction of trip interchanges between zonal pairs. While zonal destination choice models are the norm, improved computational capacity in recent times has meant that analysts can define smaller zones. Accordingly, the most spatially detailed models today operate on parcels. (A “parcel” is a land area that is managed as a single, relatively large unit, defined either to consist of a single land use or based on ownership; parcels could often, though not always, correspond to elemental destination choice alternatives.)