Modeling Tour-Based Mode Choice in Agent-Based Travel Demand Models
Year: 2026
Type: conference-paper
Abstract: Mode choice in travel demand models is commonly predicted on a trip-by-trip basis, assuming that individuals choose a mode independently for each trip. This sequential structure neglects that travelers often anticipate subsequent trips within a tour and select modes to maximize overall tour-level utility. As a result, trip-based models may generate unrealistic behavior. This paper proposes and evaluates a tour-based mode choice approach that can be integrated into an agent-based travel demand model. Using household travel survey data from northern Germany, three model variants are estimated and compared: a standard trip-based model, a trip-based model incorporating tour attributes, and a two-step tour-based model distinguishing between fixed and flexible modes. The results show that integrating tour information substantially improves behavioral plausibility. Bicycle use on long-distance tours decreases to more realistic levels, fixed-mode decisions become more consistent, and anticipatory return-trip behavior is better represented. The two-step tour-based model performs best, demonstrating clear advantages over trip-based structures. Overall, the findings indicate that tour-based mode choice modeling enhances the realism of travel demand simulations and provides a promising foundation for further extensions, including subtours and integrated destination–mode choice frameworks.
Source: Procedia Computer Science
Institution: Karlsruhe Institute of Technology
Language: English
Subfield: Transportation
Field: Social Sciences
Domain: Social Sciences
Keywords: cycling, travel demand simulation
Open Access status: diamond
OpenAlex ID: W7163195057
