基于负二项回归的水上交通事故分析与预测  

Maritime accident analysis and prediction based on negative binomial regression

作　　者：张晓东[1,2] 张浩[1] 陈亮[1] 肖英杰[1] 

机构地区：[1]上海海事大学商船学院,上海201306 [2]上海海事局,上海200086

出　　处：《上海海事大学学报》2013年第2期8-12,共5页Journal of Shanghai Maritime University

基　　金：国家自然科学基金(51149001);上海海事大学校基金(20120059)

摘　　要：寻找导致船舶损伤和事故伤亡原因的可能因素和对水上交通事故趋势预测一直是水上交通事故研究的焦点.基于事故因果关系的统计数据,分析港区航道及附近水域水上交通事故的伤亡人数与事故船舶类型、事故种类、事故发生时间、事故发生地理位置等因素之间的关系,通过比较泊松和负二项两种概率分布模型回归的结果,确定负二项分布形式的事故预测模型.研究表明,基于负二项回归的水上交通事故分析与预测方法具有适用性.It has been a focus of the maritime accident research to discover possible factors leading to ship damage and casualties and predict the trend of maritime accidents. Based on the statistical data conceming accident causality, the relationships between the casualty and the factors of maritime accidents such as the type of ship, the type of accident, the time of accident and the site of the accident are analyzed in the port channel and the waters nearby. Through comparing the results of Poisson regression model and the negative binomial regression model, the negative binomial regression model is chosen as the accident prediction model. Research shows that the negative binomial method has the applicability in analysis and forecast of maritime accidents.

关 键 词：负二项回归 泊松分布 水上交通事故 事故预测 

分 类 号：U698.6[交通运输工程—港口、海岸及近海工程]