基于AdaBoost法在代谢综合征不平衡数据分类中的应用

发布时间：2018-02-02 02:21

  本文关键词： 代谢综合征 AdaBoost 决策树 不平衡数据集　出处：《现代预防医学》2017年21期 　论文类型：期刊论文

【摘要】：目的 (1)针对医疗数据不平衡的特点,以代谢综合征为例,通过比较单纯决策树与AdaBoost+决策树分类代谢综合征的性能,从而确定AdaBoost+决策树在医疗不平衡数据挖掘中的优点,为计算机辅助诊断代谢综合征提供方法学参考。(2)采用决策树探讨代谢综合征的影响因素。方法采用AdaBoost平衡代谢综合征数据,并比较数据平衡前后决策树建模的性能,采用F-value,G-mean和AUC分析评价模型。结果 (1)相较于单纯决策树,AdaBoost+决策树的F-value值提高6.3%,G-mean提高3.5%,AUC提高0.4%,分别表明采用AdaBoost+决策树分类代谢综合征患者识别的性能提高6.3%,数据整体的分类精度提高3.5%;模型的综合分类能力提高0.4%。(2)探讨决策树影响因素均显示:空腹血糖、高密度脂蛋白、收缩压、年龄、体重指数是代谢综合征的主要影响因素。此外,在本研究中,决策树提示:若FPG6.02,BMI24.99,SBP139,age≤46,则患有代谢综合征;若FPG≤6.02,HDL-C≤0.99,BMI≤24.99,age≤61,则不患代谢综合征。结论采用AdaBoost+决策树的性能优于决策树,使用决策树所得结果与相关专业研究中代谢综合征影响因素相同。
[Abstract]:Objective 1) aiming at the characteristics of imbalance of medical data, taking metabolic syndrome as an example, the performance of simple decision tree and AdaBoost decision tree classification metabolic syndrome was compared. In order to determine the advantages of AdaBoost decision tree in medical imbalance data mining. To provide methodological reference for computer-aided diagnosis of metabolic syndrome. (2) to explore the influencing factors of metabolic syndrome by using decision tree. Methods the data of AdaBoost balanced metabolic syndrome were used. The performance of decision tree modeling before and after data balance was compared. F-valueG-mean and AUC were used to analyze and evaluate the model. F-value of AdaBoost decision tree was increased by 6.3and G-mean was increased by 3.5. the value of F-value was increased by 0.4%. The results showed that the classification performance of metabolic syndrome patients using AdaBoost decision tree was improved by 6.3 and the classification accuracy of data was improved by 3.5. The comprehensive classification ability of the model was improved by 0.4. 2) to explore the influencing factors of decision tree: fasting blood glucose, high density lipoprotein, systolic blood pressure, age. Body mass index (BMI) is the main factor affecting metabolic syndrome. In addition, in this study, the decision tree indicates that if FPG6.02BMI24.99 / SBP139age 鈮，

本文编号：1483403