Adaptive Choice Behavior of Highway Users in Response to Real-Time Road Accident Information

This study aims to investigate the adaptive choice behavior of highway users in response to real-time road accident information. The goal is to provide drivers with driving information that is more tailored to their individual needs. A large-scale SP survey was conducted, which first examined the travel information preferences of 2,500 users and then evaluated their choice behavior in different scenarios, including real-time accident information on highways and alternative road information. The study used latent class analysis to classify the 2,500 users into different clusters and analyzed the choice behavior of each cluster in response to real-time highway information. The study also explored the importance of various feature variables for different clusters using a random forest model and compared it with other machine learning algorithms. The results showed that the random forest model outperformed other models in terms of ROC curve area and accuracy. Finally, to enhance the interpretability of machine learning, the study introduced the SHAP algorithm. The results of the SHAP algorithm showed that different feature factors had significant differences in choice adaptability for different clusters of users, and choice behavior also varied in different scenarios. This study provides practical references for the development of personalized real-time traffic information systems and road traffic safety.