论文部分内容阅读
目的应用支持向量机模型法评估研究入境国际航行船舶携带输入外来医学媒介生物的风险。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834艘媒介生物阳性船舶和2 151艘媒介生物阴性船舶的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语言编程实现支持向量机模型法建模训练,并以所建模型预测新到港的1 333艘次船舶外来媒介携带风险。结果经过12种支持向量机模型的筛选,以预测精度为判定指标,选定分类器与核函数两个参数分别为“nu-classification”和“高斯函数”的支持向量机模型为最优模型,并构建起基于支持向量机的外来医学媒介生物携带风险与入境国际航行船舶关联因子间的非线性关系模型。模型训练过程的正确分类率为78.89%,通过该模型预测船舶携带外来媒介风险与实际检疫结果的符合率达到82.52%,预测效果良好。结论针对高度不确定的非线性系统,应用支持向量机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提供理论基础。
OBJECTIVE: To evaluate the risk of carrying imported exotic medical vectors by international ships entering the country of origin by using SVM model. Methods Taking Ningbo Port as the second largest port in China and the fifth largest port in the world as the research area, taking the international voyages to Hong Kong in 2014 as the research object, 33 indicators of 834 medium bio-positive vessels and 2 151 medium bio-negative vessels Expand the investigation, collecting data information. After data cleaning and variable screening, R language programming was used to implement SVM model modeling training. Based on the model built, the risk of carrying foreign carriers in newly arrived port vessels was predicted. Results After 12 SVM models were selected and the prediction accuracy was taken as the decision index, the support vector machine model with two parameters of “nu-classification ” and “Gaussian function” as the selected classifier and kernel function was Optimal model and build a non-linear relationship model between the risk of organism carrying foreign medical vectors based on support vector machine and the correlation factor of entering ships on international navigation. The correct classification rate of model training process was 78.89%. The coincidence rate of forecasting the risk of carrying foreign media and the actual quarantine results by this model reached 82.52%, and the prediction effect was good. Conclusion For highly uncertain nonlinear systems, the application of SVM model can achieve more accurate prediction function and provide a theoretical basis for the research on the assessment and early warning of the frontier health quarantine risk.