入境国际航行船舶外来病媒生物风险评估研究及应用  

Application and risk assessment of exotic medical-vector captured in the international navigation ships

作　　者：裘炯良[1] 卢厚林[1] 周力沛[1] 孙志[1] 郑剑宁[1] 施惠祥[1] 杨定波[1] 

机构地区：[1]宁波出入境检验检疫局,浙江宁波315012

出　　处：《中华卫生杀虫药械》2017年第3期230-234,共5页Chinese Journal of Hygienic Insecticides and Equipments

基　　金：国家质检总局科技计划项目(编号:2012B172)

摘　　要：目的应用随机森林模型法评估研究入境国际航行船舶携带输入外来病媒生物的风险。方法以中国第2大港、世界第5大港的宁波港作为研究范围,以2014年到港的国际航行船舶为研究对象,对6 051艘次船舶的33项指标展开调查,采集数据信息。对数据进行清洗及变量筛选后应用R语言编程实现随机森林模型法建模训练,并以所建模型预测新到港的1 333艘次船舶外来病媒生物携带风险。结果经过变量筛选,船舶总吨、净吨、船龄、货物种类等8个变量对入境国际航行船舶上是否可能携带外来病媒生物的风险预测有重要意义;最优模型是决策树节点变量数=3,决策树数量=500的随机森林模型;通过该模型预测船舶携带外来病媒生物风险与实际检疫结果的符合率达到81.17%,预测效果良好。结论针对高度不确定的非线性系统,应用随机森林模型法可实现更加精确的预测功能,为国境卫生检疫风险评估及预警方面的研究及应用提供理论基础。Objective To apply the random forests model on the risk assessment of exotic medical-vector captured in the international navigation ships based on R language.Methods Ningbo Port as the No.2 in China and No.5 in the world was selected as the researching area,and the arrival international navigation ships in 2014 were selected as the researching objects.33 indexes were surveyed for 6 051 arrival vessels. Random forests model was employed for the training and calculation. 1 333 new-arrival vessels were used for the prediction by the model. Results There were eight variables including gross tonnage, net tonnage, years old of vessel, cargoes et al related with the prediction of exotic vectors on the vessels.The model with 3 variables of nodes and 500 decision trees was concluded as the optimal prediction model. The predictive condition was good as the according rate attained 81.17%. Conclusion The relatively exact prediction is to be realized based on random forests model, especially for the highly uncertain nonlinear system.So the model can provide the theoretical basis for the risk analysis and alert of health quarantine.

关 键 词：随机森林 外来病媒生物 风险预测 R语言 大数据 

分 类 号：R184.3[医药卫生—流行病学]