A model of two-destination choice in trip chains with GPS data

Studying trip chaining behavior has been a challenging endeavor which requires the support of microscopic travel data. New insights into trip chaining can be gained from real-time GPS travel data. This research introduces a framework that considers two-destination choice in the context of home-based trip chains. We propose and empirically compare three alternatives of building choice sets where we consider various relationships of the two destinations (such as major–minor destinations, selecting one first, and selecting two concurrently). Our choice set formation alternatives use survival models to determine the selection probability of a destination. Our results reveal that trip chaining behavior is shaped by the features of retail clusters, spatial patterns of clusters, transportation networks, and the axis of travel. This research reveals that not only the spatial relationship but also the land use relationship of the destinations in a trip chain affect the decision making process.