Tour Modeling Techniques
P Vovsha, S Gupta, Parsons Brinckerhoff, US
Actual technical implementations of tour-based models are very different and the methods are still evolving. The paper summarizes the existing and proposed tour modeling techniques, and identifies their relative advantages and disadvantages.
Considering tour instead of trip as the main unit for travel modeling is the cornerstone of Activity-Based Models (ABMs). Major theoretical and practical advantages of tour relate to the interdependences of mode, destination, and time-of-day choices across different trips on the same tour. The actual technical implementations of tour-based models are very different and the methods are still evolving. The main problem is that it is impossible to model all destinations, modes, and departure times simultaneously for a multi-destination tour because of the infeasible number of possible combinations.
The paper summarizes the existing and proposed tour modeling techniques, identifies their relative advantages and disadvantages, as well as directions for future research. The methods are grouped into three broad categories:
1. Rubber band. This approach is the most frequent in applied travel ABMs in the US. First, tours are generated for each individual by purpose of the primary destination. Depending on the model system details, tours might be generated with or without intermediate stops. Then, for each tour, primary destination location, time-of-day choice, and tour mode combination are modeled. Finally, intermediate stops are inserted based on the known frequency or by means of a stop-frequency choice model. In this process, location for each stop, trip mode, and departure time are defined. The approach received its name from the way how intermediate stop are located, all else being equal, inversely proportional to the deviation from the shortest path between the home and primary destination. While this approach has been widely adopted in practice, it has been criticized for a lack of integrity between location and mode choices at the tour and trip level. Also, the very fact that tours rather than activities are generated from the very beginning is recognized as a drawback.
2. Incremental tour construction. This approach has been applied in both travel ABMs and freight models. Tour purposes and primary destinations are not identified from the very beginning. The tour structure rather emerges from the sequence of choices for each subsequent activity (departure time from the previous activity, location, and mode). Tour starts when the person whose previous activity was at home chooses the next activity out of home. Tour ends when the person whose previous activity was out of home chooses the next activity at home. This approach is especially appealing for commercial deliveries and shopping travel tours where identification of the primary destination is quite arbitrary. However, in many other travel contexts, this approach looks problematic, and primarily because of the sequential chronological order of processing activities. This approach is not applicable to tours where the primary destination is clearly identified (majority of work, school, and discretionary tours) and might not necessarily be the first activity chronologically.
3. Combinatorial tour formation. In the genuine ABM paradigm, activities should be modeled first and tours should be formed as a means to reach those activities. It is proposed to distinguish between three types of activities: 1) Activities with a fixed location and schedule (start and end time). These activities are not bound to work and school but also include doctor/dentist appointments, shows, sport events, dropping-off and picking-up passengers, etc. It is assumed that individuals build their daily schedules and travel tours pivoting off these activities; 2) Activities with a fixed location but flexible schedule like visiting a certain store for a particular shopping type. It is assumed that individuals try to link these activities to the other activities in order to optimize their travel arrangements and capitalize on the schedule flexibility. It can be assumed that most of travel tour skeletons are defined by the first two types of activities; 3) Activities with a flexible location and schedule like grocery shopping or visiting a coffee shop. These activities are most frequently "inserted" in the travel tours depending on the time-space constraints but rarely play a role of the primary destination (except for short non-motorized tours). A combinatorial tour formation model is proposed that address the major activities of first two types while the travel-driven activities should be inserted as additional stops on the tours (in the way similar to the rubber band approach).
Association for European Transport