论文部分内容阅读
为分析船舶溢油事故报告行为的影响因子和偏好,构建以环境因子和事故船方主体特征为自变量、报告选择为因变量的关系结构,从连通性、效率和效果3个角度完成了相关信息采集。基于调查问卷数据,分别利用Probit模型和贝叶斯网推理明确了影响事故报告行为的关键因子,基于ROC值对两个模型的预测精确度进行了评估,选择精确度较高的模型预测事故报告行为偏好。结果表明,在船工龄、职务、培训情况、溢油数量和事故时间会对溢油事故报告的具体行为偏好产生影响。
In order to analyze the influencing factors and preferences of the reporting of ship oil spill accidents and to construct the relational structure between environmental factors and the main characteristics of the accident ship as independent variables and the selection of reporting variables as dependent variables, the correlation is completed from the perspectives of connectivity, efficiency and effectiveness Information Collection. Based on the questionnaire data, the key factors influencing the accident reporting behavior are identified respectively by Probit model and Bayesian network reasoning. The prediction accuracy of the two models is evaluated based on the ROC value, and the model with higher accuracy is selected to predict the accident report Behavior preferences. The results show that the length of service, positions, training, number of spills and accident times have an impact on the specific behavioral preferences of oil spill incidents.