An accident severity prediction framework with consideration of features interaction

Accident severity prediction is increasingly important for preventing and reducing losses due to traffic accidents. However, many studies ignore the complex relationship between features during feature selection. To improve the prediction accuracy of an accident severity prediction model, in this paper, the interactions between multiple features are considered. First, the feature selection algorithm of recursive feature elimination with cross-validation is improved by using Shapley additive explanations as the feature importance assessment metric. Then, to decrease the time expense of manually finding hyperparameters of the model, the hunter–prey optimisation (HPO) algorithm is introduced and logistic mapping together with stochastic perturbation is added to it, which makes it easier to skip out of the partial optimum during the optimisation search. Finally, the improved HPO algorithm is used to optimise the hyperparameters of the CatBoost model. The US traffic accident dataset is introduced for the validity of the proposed framework. Experimental results show that the proposed framework achieves a prediction accuracy of 96.63%, which is better than other state-of-the-art methods. The high accuracy of the prediction model can help decision-makers develop more rational transportation policies; some traffic management measures are also proposed in this study, based on the selected features.