机动车与非机动车事故严重度多因素影响分析

Analysis of the multifactorial influence of motor vehicle and non-motor vehicle accident severity

为探讨机动车与非机动车事故严重度的多因素影响情况,对某市的1304起真实机动车与非机动车碰撞事故数据进行研究。采用列联分析方法对事故严重度相关因素进行显著性分析,发现9类因素与事故严重度显著相关;建立C5.0决策树模型,对所提取的显著性因素进行关联性分析,根据混淆矩阵对分类进行评价并与未引入列联分析方法的分类结果进行对比;通过提取模型中的决策规则获得机动车与非机动车死亡事故多因素诱发路径。结果表明,引入列联分析避免了因掺杂对事故严重度无显著性影响的因素而引发的错误分类结果,且机动车类型是影响事故严重度的关键因素;而非机动车骑行者受教育程度、机动车与非机动车类型和时段因素是死亡事故的重要影响因素。

In order to investigate the multifactorial influence of motor vehicle and non-motor vehicle accident severity,the data of 1304 actual motor vehicle and non-motor vehicle collision accidents in a city were analyzed.Firstly,crosstabs analysis was used to determine the significant factors correlated to motor vehicle-non-motor vehicle accident severity.It was found that nine factors were significantly associated with accident severity.Secondly,on this basis,the extracted significant factors were analyzed for correlation by building the C5.0 decision tree model,and the classification results were not only evaluated based on the confusion matrix,but also compared with the classification results without using crosstabs analysis.Finally,the multi-factor predisposition paths for motor vehicle and non-motor vehicle fatal accidents were obtained by decision rules.The results indicate that the introduction of crosstabs analysis avoids misclassification results due to adulteration with factors that have no significant effect on accident severity,and the key factor affecting accident severity is the type of motor vehicle;moreover,the non-motorized rider′s education,motorized and non-motorized vehicle types,and time period are main factors of fatal accidents.