| To alleviate traffic congestion without building new facilities or expanding current systems, advanced traveler information systems (ATIS) have been studied and implemented during the past several decades. Given the continued increase in the demand for mobility and safety of travelers under limited opportunities for physical expansion of the system, ATIS are essential tools for managing and controlling traffic. However, the performance of ATIS depends heavily on the positive response of drivers to the system. Thus, to enhance the systems, it is very important to evaluate and understand driver behavior characteristics with ATIS.; This study explored many details of the driver's response to variable message signs (VMS) which is the one of the most common ATIS deployed in many areas. Three mail-back surveys and an on-line survey were conducted in Wisconsin to collect various drivers' behavior information with VMS. Based on the findings from the surveys, seven potential affecting factors for driver compliance with ATIS were identified and applied to further study. A driver compliance survey was conducted to evaluate the significance of these seven factors.; Four different approaches: binary logistic regression, artificial neural network (ANN), classification trees (QUEST), and logistic regression trees (LOTUS) were adopted to develop the driver compliance model. Anticipated travel time savings, knowledge of an alternate route, and level of confidence in the alternate route proved to be significant variables affecting driver compliance with VMS. The LOTUS model revealed the hierarchical structure of driver compliance and produced many interesting findings. The developed models were evaluated with newly collected data from on-line survey. All four models showed reasonable performance, while the LOTUS model explained the behavior data most clearly and the ANN model maintained the highest prediction rate. |
