零膨胀泊松回归模型及其在交通事故中的应用  

Zero-inflated Poisson Regression Model and Its Application in Traffic Accident

作　　者：陈异[1] 戴琳[1] 寇鹏[1] 

机构地区：[1]昆明理工大学理学院,云南昆明650093

出　　处：《计算机技术与发展》2013年第10期163-166,共4页Computer Technology and Development

基　　金：国家自然科学青年基金(11201200);国家自然科学天元基金(11126310);云南省基金(2010CZ059)

摘　　要：零点膨胀泊松回归模型是利用零处的概率质量退化分布和一个泊松回归模型进行混合所得到的,该模型是分析零点膨胀计数数据的有效工具。文中采用零膨胀泊松回归模型对某高速公路数据进行拟合并对该数据采用score检验统计量就是否存在零膨胀进行了检验,研究结果表明当车流量达到1.8辆万/日以上交通事故发生的频率会明显增加,有关部门应采取相应的措施,如限制收费站进口车辆或者采取分流等,特别是在极端天气出现时,更应及时控制好车流量并提醒驾驶员保持车距。Zero-inflated Poisson regression modeluses the zero probability mass distribution and a Poisson regression model for mixing. The model is very effective to analyze and research the problem of excessive containing zero. In this paper,adopt the zero-inflated Pois-son regression model to research and analyze the relationship about highway traffic flow and traffic accident frequency,and use score test statistic for test whether there is zero-inflation existed in this count data,obtaining the zero-inflated model is valid. Research results show that when the traffic flow reached 1. 8 million/day or more traffic accident frequencies will increase obviously,the relevant departments should take the corresponding measures,such as restrictions on imported vehicle toll station or adopting billabong,especially in extreme weather occurs,it should prompt control car traffic and remind the driver to keep distance.

关 键 词：零膨胀 ZIP模型 SCORE检验 交通事故 

分 类 号：O212.4[理学—概率论与数理统计]