基于判别分析的骑推行交通事故鉴定研究

发布时间：2018-04-02 19:33

  本文选题：交通事故　切入点：司法鉴定　出处：《数学的实践与认识》2017年08期

【摘要】：道路交通事故鉴定对于交通事故责任认定和法庭举证具有重要作用.采集某市近三年125起交通事故案例数据,建立机动车、自行车、人体总计73个调查变量,运用SPSS软件开展骑推行事故鉴定研究.通过相关性分析发现车座、机动车类型等6个变量与鉴定结论相关.开展线性判别分析研究,结果表明直接对相关变量进行判别分析可以快速获得最佳判别效果.当函数选入的自变量为车座时,交叉验证准确率最高可达72.8%,说明采用数据挖掘的方法来鉴别交通事故中的行为方式具有一定可行性.
[Abstract]:Road traffic accident identification plays an important role in the identification of traffic accident liability and court proof.The data of 125 traffic accidents in a certain city in recent three years were collected, and a total of 73 investigation variables of motor vehicles, bicycles and human bodies were established, and the identification of riding accidents was carried out by using SPSS software.Through the correlation analysis, it is found that six variables, such as vehicle seat and motor vehicle type, are related to the identification conclusion.The results of linear discriminant analysis (LDA) show that the best discriminant effect can be obtained by direct discriminant analysis of related variables.When the independent variable selected by the function is the seat, the accuracy of cross-validation can reach 72.8, which shows that it is feasible to use data mining method to identify the behavior in traffic accidents.
【作者单位】： 北京工业大学北京市城市交通运行保障研究中心;天津职业技术师范大学天津市交通安全与控制协同创新中心;北京市交通信息中心;
【基金】：国家自然科学基金(61301040) 天津市自然科学基金重点项目(16JCZDJC38200) 天津职业技术师范大学校级教学改革与质量建设研究项目(JGY2015-15)
【分类号】：U491.31
，

本文编号：1701800