Modeling Incident-Related Routing Decisions by Using a Nested Logit Structure

Abstract

Unusual congestion that could be caused by an incident or other traffic problems is a major source of delay for drivers in urban areas. Real-time traffic information, the building block for advanced traveler information systems (ATIS), has a promising potential for alleviating such congestion by encouraging and assisting drivers to divert to less congested routes. Traffic information is envisioned to help more informed routing decisions in case of incident-related congestion. Drivers’ routing decisions made when they are faced with such unusual congestion are investigated. The factors that influence these decisions are explored, including the effect of traffic information. A nested logit modeling structure is introduced. This model proved that the nested logit approach is superior than the simple multinomial logit in modeling the choice in cases of incident-related congestion. The model also showed that the decisions not to divert from the usual route and to divert but only around the location of the problem share unobserved terms. Familiarity and usual use of alternative routes did not affect the decision in the case of an incident. Drivers who use more than one route to work do not necessarily switch routes if they encounter unusual congestion. The nested logit model also proved the significance of traffic information, indicating a promising potential benefit of ATIS in alleviating nonrecurring congestion.