TRAVEL TIME PREDICTION AND INFORMATION AVAILABILITY IN COMMUTER BEHAVIOR DYNAMICS

Abstract

The prediction of travel time by trip makers constitutes an important component of the complex daily dynamics of commuter behavior, which are of particular concern in systems evolving towards equilibrium, such as after major traffic control changes or disruptions due to major reconstruction or maintenance activities. The day-to-day dynamics of the prediction of travel time by commuters on their next trip, with particular emphasis on the effect of information availability, are investigated in this paper using an experimental approach involving commuters in a simulated commuting system. A travel time prediction model developed previously for a limited information situation provides the framework for analyzing this phenomenon, using results obtained from a second experiment where users are provided with complete information on the previous day's performance. Insights into the effect of information availability are obtained through the comparative analysis of the model's performance and estimated parameter values in the two experiments. The results suggest that additional information tends to reduce the perceived uncertainty associated with the system's performance; commuters combine this supplied information with their latest experienced travel time in forming a base value for the predicted travel time on the next trip. This base value is adjusted by a safety margin that is primarily governed by the latest experienced schedule delay, in order to protect against unacceptably late or early arrival at the workplace.