Real-time crash risk prediction in freeway tunnels considering features interaction and unobserved heterogeneity: A two-stage deep learning modeling framework
Published in Analytic Methods in Accident Research, 2023
Recommended citation: Jieling Jin, Helai Huang, Chen Yuan, Ye Li, Guoqing Zou, & Hongli Xue (2023) Real-time crash risk prediction in freeway tunnels considering features interaction and unobserved heterogeneity: A two-stage deep learning modeling framework , Analytic Methods in Accident Research, DOI: https://doi.org/10.1016/j.amar.2023.100306 https://www.sciencedirect.com/science/article/pii/S2213665723000416
– Designed a two-stage framework blending statistical and deep learning models for real-time crash risk prediction in freeway tunnels which considering features interaction and unobserved heterogeneity.
– Explored the unobserved heterogeneity and influence mechanism of precursors on real-time crash risk by a random parameters logit model with heterogeneity in means and variances.
– Developed a random deep & cross network model considering feature interactions and unobserved heterogeneities to predict real-time crash risk.
– The random deep & cross network model is explained by the shapley additive explanations approach, which discusses the mechanisms influencing the risk of tunnel crashes as well as the interactions and unobserved heterogeneity of features in the model.
Recommended citation: Jieling Jin, Helai Huang, Chen Yuan, Ye Li, Guoqing Zou, & Hongli Xue (2023) Real-time crash risk prediction in freeway tunnels considering features interaction and unobserved heterogeneity: A two-stage deep learning modeling framework , Analytic Methods in Accident Research, DOI: https://doi.org/10.1016/j.amar.2023.100306.