Scholarlink

<p>Farm Equipment Vehicles (FEVs) pose major safety challenges on interurban and rural roads, especially in developing countries where outdated fleets and poor infrastructure are common. Interactions among multiple risk factors often increase crash severity. This study investigates the severity of FEV-involved crashes and evaluates models for predicting accident outcomes with greater accuracy for Intelligent Transportation Systems (ITS). Two advanced methods are applied: a Random Parameters Ordered Logit (RPOL) model to capture observed and unobserved heterogeneity, and a Multi-Class Support Vector Machine (MC-SVM) to benchmark predictive performance. The RPOL results highlight key factors influencing crash severity, including overturns, time of day, road type, and involvement of other vehicles. Motorcycle involvement and nighttime crashes significantly raise fatality risks, while rear-end collisions and reversing maneuvers are linked to lower severity. Heterogeneity is observed in collisions where FEVs are at fault, particularly on straight roads. The MC-SVM model demonstrates superior predictive accuracy, achieving higher Area Under the Curve (AUC) scores compared to RPOL. Policy implications include renewing FEV fleets, restricting older vehicles, and imposing stricter controls on nighttime operations. Integrating statistical and machine learning approaches provides a robust framework for improving safety and reducing crash severity in rural transportation systems.</p>

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

Modeling Farm Equipment Vehicle crash injury severity using random parameters logit and multi-class support vector machine in a developing country

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月4日

查看全文
  • Ehsan Jafari Nasab,
  • Abdoreza Sheikholeslami,
  • Jose Manuel Vassallo,
  • Amin Moeinaddini

摘要

Farm Equipment Vehicles (FEVs) pose major safety challenges on interurban and rural roads, especially in developing countries where outdated fleets and poor infrastructure are common. Interactions among multiple risk factors often increase crash severity. This study investigates the severity of FEV-involved crashes and evaluates models for predicting accident outcomes with greater accuracy for Intelligent Transportation Systems (ITS). Two advanced methods are applied: a Random Parameters Ordered Logit (RPOL) model to capture observed and unobserved heterogeneity, and a Multi-Class Support Vector Machine (MC-SVM) to benchmark predictive performance. The RPOL results highlight key factors influencing crash severity, including overturns, time of day, road type, and involvement of other vehicles. Motorcycle involvement and nighttime crashes significantly raise fatality risks, while rear-end collisions and reversing maneuvers are linked to lower severity. Heterogeneity is observed in collisions where FEVs are at fault, particularly on straight roads. The MC-SVM model demonstrates superior predictive accuracy, achieving higher Area Under the Curve (AUC) scores compared to RPOL. Policy implications include renewing FEV fleets, restricting older vehicles, and imposing stricter controls on nighttime operations. Integrating statistical and machine learning approaches provides a robust framework for improving safety and reducing crash severity in rural transportation systems.