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.
Road traffic accidents Identification has an important role for responsibility iden- tification and forensic evidence. Collecting 125 cases of traffic accidents data of a city in the past three years,a total of 73 survey variables about motor vehicles, bicycles and human were established to research the forensic appraisal of wheeling or riding with SPSS software. Six variables, such as bicycle saddle and vehicle type, were related to expert conclusion. The result shows that applying linear discrimiuant analysis to handle the six variables can get the best identification effect quickly. The cross validation accuracy rate reached 72.8% when the independent variable of the function is bicycle saddle.it shows that the data mining method to identify traffic accident behavior has certain feasibility. The method of data mining has a certain feasibility to identify the manner of action in road traffic accidents.
出处
《数学的实践与认识》
北大核心
2017年第8期215-221,共7页
Mathematics in Practice and Theory
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)
关键词
交通事故
司法鉴定
判别分析
骑推行
traffic accidents
forensic appraisal
discriminant analysis
wheeling or riding
