Studies on accident severity on mountainous freeways have predominantly centered on the personal injury level, rather than the aggregation level. However, for quantifying the accident causality, clustering the accident severity from multidimensional perspectives based on data-driven approach is seldom investigated in existing studies. To address this research gap, we propose a two-stage methodology that integrates accident clustering with Bayesian inference. Initially, a Gaussian mixture clustering algorithm is developed to categorize accident severity. Subsequently, a Bayesian network is constructed to explore the risk factors associated with accident severity. The proposed model is calibrated and validated using accident data collected from mountainous freeways in Yunnan Province, China, spanning the period from 2016 to 2021. The findings suggest that our proposed accident clustering method exhibits superior robustness compared to alternative clustering techniques. Bayesian inference analysis further elucidates that accident severity is significantly influenced by factors such as driving behavior, weather conditions, and road surface conditions.
Exploring the Causality of Accident Severity on Mountainous Freeways With a Two-Stage Approach
2025
Article (Journal)
Electronic Resource
Unknown
Metadata by DOAJ is licensed under CC BY-SA 1.0
Research on Traffic Accident Prediction Model for Mountainous Freeways
British Library Conference Proceedings | 2010
|Statistical analysis of accident severity on rural freeways
Online Contents | 1996
|Springer Verlag | 2019
|