This study applies the random parameter logit model (RPL), random parameter logit model with mean and variance heterogeneity (RPLMV), random forest (RF), and extreme random tree (ERT) to investigate the factors influencing the severity of both freeway and non-freeway crashes. Furthermore, it aims to analyze the disparities between logit models and tree-based machine learning models in terms of their predictive performance and interpretability. The findings of this research contribute to a better comprehension of the disparities in prediction accuracy and interpretive capabilities between logit models and machine learning models. Moreover, they provide insights into the selection and application of computational SHAP methods. Additionally, the outcomes can serve as valuable references for governmental bodies and organizations in formulating policies.
Analysis of injury severity in freeway and non-freeway crashes based on logit model and machine learning
2023-08-04
1124151 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Classifying features of freeway crashes using machine learning
Taylor & Francis Verlag | 2022
|Analysis of Crashes on Freeway Weaving Sections
ASCE | 2020
|