Understanding and Modeling Drivers’ Diversion Behavior during Congestion

Abstract

Traffic congestion causes significant economic losses due to delays and excessive fuel consumption. Understanding drivers’ behaviors, particularly in terms of diversionary routing to avoid congestion, is crucial for addressing this issue. This study investigates driver behavior during congestion and develops a predictive model for route diversion decisions using about 20 million anonymized trips from global positioning system (GPS) data in Sydney, Australia. Among the most unique and significant contributions of this work was the development of a methodology to identify and graphically depict commute trips and assess the impact of prevailing traffic conditions and related driving factors on diversion likelihood without relying on access to supplemental data. Examples of driving factors considered include distances and times from origins and destinations, durations of delays caused by congestion, length of congested areas, and roadway classification. Findings indicate that experienced delay per congestion distance and remaining distance to the destination positively influence diversion, while expected increases in travel time on alternative routes discourage it. These results can enhance traffic simulation models and improve traffic management strategies. Overall, the contributions of this paper have both practical importance and theoretical applicability and provide a long-absent step toward understanding how individual drivers respond to congestion and make subsequent routing choices.