交通工学研究発表会論文集
Online ISSN : 2758-3635
The 43rd Conference of Japan Society of Traffic Engineers
会議情報

第 43 回 交通工学研究発表会
Exploring Contributing Factors to Accident Severity Based on Random Forest Approach
Xingwei LIUJian XINGFumihiro ITOSHIMAKuniaki SASAKI
著者情報
会議録・要旨集 認証あり

p. 89-95

詳細
抄録

Traffic accidents have grave implicationsin terms of human life and property. Efficient traffic management requires a profound comprehension of the underlying causes of accidents and the ability to predict their severity partially. In this study, we investigated the factors contributing to accident severity by utilizing accident data collected from the Gotenba to Tokyo section of the Tomei Expressway in Japan during 2019. We employed a random forest model on the cleansed dataset to predict traffic accident severity, encompassing a total of 701 cases. Additionally, a grid search was conducted to identify the optimal hyperparameters for this random forest model. To gain the independent performance and impact of each factor on traffic accident severity, we employed SHAP (SHapley Additive exPlanations) to show the visualization results. This effective tool facilitated the identification of high-risk routes and individuals. Notably, our analysis revealed that accidents occurring at the end of congestion were more prone to severity. These compelling findings provide valuable insights for the development of strategies aimed at enhancing expressway management.

著者関連情報
© 2023 Japan Society of Traffic Engineers
前の記事 次の記事
feedback
Top