中文摘要：

运用零膨胀负二项回归模型,分析水上交通事故中人员死亡失踪数量的各影响因素的影响程度。通过Pearson相关分析得到与水上交通事故中人员死亡失踪人数密切相关的因素,建立ZINB回归模型得到各因素的参数估计值。运用弹性分析方法对不同影响因素的影响程度进行评估,根据影响程度对各影响因素进行排序。将该方法应用于长江海事局辖区范围内的水上交通事故人员死亡失踪分析,取得了理想效果。

英文摘要：

To reduce the number of casuahies caused by the maritime traffic and transportation accidents in the Yangtze River reach and develop the corresponding management measures, the paper has introduced a method known as the zero-inflated negative binomial regression to analyze the influential factors and degrees of such factors influencing the maritime traffic and transportation accidents based on the data collection of such accidents that took place in the reach during the period of the last seven years by using the regression model of ZINB and that of ZIP. The model testing and investigation has made us gain the consequential out- comes of the order by using a regression model of ZINB through careful and close observation of the data gained with the regression model of ZIP. In the paper, we have gained the related factors through the Pearson correlation analysis and established the parameter estimation approach to the ZINB model, thus, determining the influential degrees of each factor through the elastic analysis. And, then, the influential factors have been rationally categorized into the related influential degrees. The results of the analysis indicate that, the influential factors that cause the maritime traffic and transportation accidents should be kept away in accord with the degree of influence on the number of missing deaths in the fire/explosion, grounding collision, contact damage, the sinking and stranding, in addition to bad weather hitting casualty. Among them, the fire/explosion proves much more threatening in leading to such human casualties or missing deaths in such maritime accidents. Furthermore, the results of the elastic statistics have also shown that, the more passengers the shipping transportation and traffic have dealt with and the greater the size of the maintenance dimensions of the channel are involved, the greater number of human casualty （including the passenger missing） there will be. Therefore, we have proposed that it is necessary to use the analysis method for the statistics of