Scholarlink

Road safety represents a pivotal global challenge, with traffic accidents precipitating substantial burdens on public health, the economy, and society at large. In particular, inexperienced drivers are susceptible to risk, particularly in conditions of low visibility. This study addresses the existing knowledge gap in understanding the impact of rainy nights on accident severity by considering factors such as throttle, brake, and wheel positions. A tree-based machine learning framework (Decision Tree, Random Forest, Gradient Boosting, Extra Tree Classifier) was employed to predict accident severity. To overcome the imbalanced dataset, a rigorous feature selection and data balancing techniques were implemented. The findings demonstrate the superior performance of the Random Forest algorithm in predicting accident severity compared to other Tree based algorithms. This model exhibited high accuracy (84%) in identifying accident events, outperforming the results of the state-of-the-art studies (70%), which highlights its potential for improving road safety measures.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Advanced Tree-Based Ensemble Learning System for Prediction Accident Severity with Imbalanced Data Handling

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Naima Goubraim,
  • Zouhair Elamrani Abou Elassad,
  • Naima El-Amarty,
  • Hajar Mousannif

摘要

Road safety represents a pivotal global challenge, with traffic accidents precipitating substantial burdens on public health, the economy, and society at large. In particular, inexperienced drivers are susceptible to risk, particularly in conditions of low visibility. This study addresses the existing knowledge gap in understanding the impact of rainy nights on accident severity by considering factors such as throttle, brake, and wheel positions. A tree-based machine learning framework (Decision Tree, Random Forest, Gradient Boosting, Extra Tree Classifier) was employed to predict accident severity. To overcome the imbalanced dataset, a rigorous feature selection and data balancing techniques were implemented. The findings demonstrate the superior performance of the Random Forest algorithm in predicting accident severity compared to other Tree based algorithms. This model exhibited high accuracy (84%) in identifying accident events, outperforming the results of the state-of-the-art studies (70%), which highlights its potential for improving road safety measures.