【作者】 李霞； 何丽云； 刘超；

【Author】 LiXia,He LiYun,LiuChao,School of Statistics Renmin University of China,China Academy of Traditional Chinese Medicine

【机构】 School of Statistics Renmin University of China； China Academy of Traditional Chinese Medicine；

【摘要】 目的:应用Boosting算法建模,对亚健康状态的人群进行分类并分析其临床特征,同时与通常使用的logistic统计方法进行比较研究。方法:使用不同损失函数下boosting算法对亚健康状态流行病学调查数据的健康和亚健康人群进行分类;同时,通过重要性度量分析得到亚健康状态的重要临床特征。结果:采用boosting算法对亚健康人群进行了分类得到了分类准确性及重要影响变量。结论:使用boosting算法对亚健康人群进行分类的方法比传统的两种logistic回归模型分类准确率要高,同时也可以得到影响亚健康状态的重要变量。更多还原

【Abstract】 Objective The purpose of this study was to apply boosting algorithm to construct a model in"sub-health"epidemiological survey data,finally get the classification model and the clinic characteristic of"sub-health"crowd.Methods Using boosting algorithm with different loss functions to classify the"sub-health"data,and use the important measure to get the clinic characteristic.Results We got the classification accuracy of the final model and the clinic characteristic, our model had superiority to classical logistic regression model.更多还原