为制定中国青年女性红细胞压积参考值的统一标准提供科学依据,收集中国各地用温氏法测定的健康青年女性红细胞压积参考值,提出了一种基于支持向量机的分析青年女性红细胞压积参考值与地理环境的海拔高度、年日照时数、年平均相对湿度、年平均气温、年降水量等要素关系的方法.知道了中国某地的地理因素,就可以用此模型预测这个地区的青年女性红细胞压积参考值.实验结果表明,支持向量机预测模型能够克服地理要素本身之间具有的共线性,反映红细胞压积参考值与地理环境之间的非线性、复杂性关系特征,并且保持一定的预测精度,与主成分回归模型相比,更适应于红细胞压积参考值的预测.
In order to provide a basis for unifying the reference value criteria standard of Chinese young women's hematocrit,the paper discusses the nonlinear relationship between the reference value of Chinese healthy young women's hematocrit which are determined by the wintrobe laws and geography factors based on support vector machines.The selected geographical factors include altitude,annual sunshine hour,annual average relative humidity,annual average temperature and annual precipitation.If the geographical values are obtained in some area,the reference value of Chinese young women's hematocrit of this area can be reckoned using this SVM model.Experimental results show that the SVM model is capable of overcoming the multicollinearity between the geography factors and maintaining the stability of the predictive accuracy,and is more suitable for predicting the hematocrit value than the principal component regression analysis.