This paper proposes a machine learning method based on a two-layer Stacking ensemble learning strategy to predict the severity of traffic accidents. In the first layer, the prediction results obtained by learning from three base learners, namely KNN, XGBoost, and LightGBM, are combined and input as new features into the second-layer meta-learner LR, thus establishing a combined prediction model. The experiment selects traffic accident data from the UK for verification and analysis. The results show that this prediction model has more superior prediction accuracy and generalization performance compared to other single models. The prediction accuracy of the traffic accident severity on the test set reaches 0.889, the AUC value is 0.905, and it outperforms other models in terms of precision, recall, and F1 value. This research can assist traffic management departments in preventing traffic accidents, thereby reducing the severity of traffic accidents and decreasing the occurrence of traffic accidents.
Ensemble learning-based prediction of road traffic accident severity
International Conference on Smart Transportation and City Engineering (STCE 2024) ; 2024 ; Chongqing, China
Proc. SPIE ; 13575
2025-04-28
Conference paper
Electronic Resource
English
Prediction of road traffic accident severity based on multi-model fusion
British Library Conference Proceedings | 2023
|Traffic accident severity prediction based on interpretable deep learning model
Taylor & Francis Verlag | 2025
|