With the rapid development of urban transportation, the number of traffic incidents has increased year by year. Traffic incidents not only cause serious damage to the lives and properties of travelers but also become a great challenge to traffic operation and management. Particularly, the duration of traffic incidents is an important indicator to measure the impact of incidents. Many previous studies have focused on the duration of traffic incidents on the freeway rather than the urban roads. Moreover, the time range and research area are often limited. In our study, we use the traffic incident data of Tianjin, China in two consecutive years. We utilize two regression models and two tree-based models to analyze and predict the duration of traffic incidents. The results show that the overall prediction accuracy of the log-linear regression model is higher than the other three approaches. Meanwhile, factors like “whether in the urban area”, “road type”, “incident type” and “number of people involved” have a significant influence on the duration.

CICTP