Two basic options exist for modeling transit path-choice in the travel forecasting process: including transit path-choice in a detailed mode choice model or using deterministic procedures to model transit path-choice through transit multipath, path-building algorithms. These two options require different approaches to transit path-building, mode choice, and transit assignment. Modeling decisions regarding transit path-building techniques and the amount of transit path-choice incorporated into mode choice models directly impact the ability of travel models to consistently model mode choice and transit use. This paper discusses different options for transit path-building and mode choice and the implications of those options. It then presents an analysis of the success of alternate transit path-building techniques in reproducing known transit paths as reported by transit users in Denver as well as an analysis of several commonly used transit path-building validation techniques. The paper provides several suggestions regarding the interface transit path-choice and transit path-building based on the results of the empirical transit path-building tests. Two primary suggestions applicable to current planning processes have been made. First, transit path-choice should be performed in only one component of the forecasting process such as transit path-building or mode choice in order to minimize inconsistencies in the modeling process. Second, disaggregate measures such as prediction success tables comparing modeled boardings to reported boardings for individual observations of transit trips should be compiled for validation of transit networks and transit path-building procedures.
Samenvatting