Search

Efficient path guidance is one of the effective ways to improve the utilization of road resources and relieve traffic congestion. Therefore, it is important to identify ways to enable drivers to select path more efficiently. In this paper, we seek to build a pre + during-trip path prediction model based on the impacts of external and internal information on path selection behaviors to enhance path selection efficiency. The prediction process is composed of three parts, namely, pre-trip path prediction, during-trip link prediction, and path adjustment calculation. In the pre-trip path prediction, through RP survey data collected in Changchun, China, we determine the impacts of subjective factor (habits) on path selection using the Markov model. While the Binary Logit model is utilized by the during-trip link prediction to consider the impacts of both subjective factor (travel habits) and objective factor (real-time traffic conditions). The influences of habits and traffic conditions are compared and analyzed. The results indicate that subjective factor has more important influence than objective one. In addition, the verification results also suggest that pre + during-trip path prediction model provides higher forecasting accuracy than the single pre-trip prediction model. These findings are beneficial to uncover the underlying mechanisms of path selection and facilitate the development of strategies to enhance path selection efficiency. Based on the study results, the subjective and objective information that was found to affect path selection can be considered into the path guide system. Moreover, the hybrid prediction model can be applied in the vehicle or mobile navigation App to facilitate the recommendation of the path to travelers as well as to forecast the short-term traffic flow and determine the potential congestion area. Based on the pre-trip path prediction model, the path and the traffic flow distribution in road network can be obtained only from historical travel data.