交通事故人员伤亡的影响因素分析  

Analysis of factors influencing traffic accident casualties

作　　者：李蕊[1] 赵丽华[1] 

机构地区：[1]太原理工大学数学学院,山西太原030024

出　　处：《中国安全科学学报》2015年第4期123-127,共5页China Safety Science Journal

基　　金：山西省青年科技研究基金资助(2012021015-6)

摘　　要：为探究交通事故各影响因素对人员伤亡的影响,并根据影响情况给出减少事故伤亡的合理建议,通过收集、统计某市2013年4—9月的交通事故数据,采用Poisson、负二项(NB)、HurdlePoisson(HP)、Hurdle-NB(HNB)、零膨胀Poisson(ZIP)和零膨胀负二项(ZINB)回归模型,对交通事故计数资料进行回归拟合,通过似然比(LR)检验、Vuong检验和拟合优度检验准则比较与选择模型。结果显示,ZINB回归模型拟合效果最佳,肇事车辆有无牌照、车损情况、驾驶人驾驶资历和天气状况等因素对事故中伤亡人数的影响显著,采取限制无牌照车辆出行,保障车辆性能和尽量让驾驶资历高的人驾驶并控制车速等措施能够减少事故伤亡。Various factors influencing traffic accidents casualties, were analyzed on the basis of traffic accident data obtained for a certain city from April to September in 2013. The data were fitted respectively using Poisson, negative binomial （NB）, hurdle-Poisson （HP）, hurdle-NB （HNB）, zero-inflated Poisson （ZIP） and zero-inflated negative binomial （ZINB） regression models. LR test, Vuong test and goodness- of-fit test criterion were used to evaluate the models . The results show that ZINB regression model fits the data better than the other models, that having the vehicle license, vehicle damage , the driver＇s driving qualification and weather condition and other factors have a significant effect on the number of casualties, and that he number of casualties can be reduced by controlling the vehicle without license, ensuring the vehicle performance, and limiting vehicle velocity among other measures.

关 键 词：交通事故 零膨胀 Hurdle模型 似然比(LR)检验 Vuong检验 

分 类 号：X928.03[环境科学与工程—安全科学]