基于信息增益率的超高维变量选择  

Ultra-high Dimensional Variable Selection Based on Information Gain Ratio

作　　者：朱燚丹 陈兴荣[1] 李秋萍 Zhu Yidan;Chen Xingrong;Li Qiuping(School of Mathematics and Physics,China University of Geosciences(Wuhan),Wuhan 430074,China)

机构地区：[1]中国地质大学(武汉)数学与物理学院,武汉430074

出　　处：《统计与决策》2021年第22期18-21,共4页Statistics & Decision

摘　　要：文章针对信息增益变量选择方法(IG-SIS)的不足,对其进行改进,提出适用于超高维、无模型假设框架下基于信息增益率的变量选择方法(IGR-SIS),从理论上证明了IGR-SIS方法具有确定性筛选性质,并通过蒙特卡洛数值模拟和基因表达分类数据验证IGR-SIS方法对超高维分类变量选择的有效性。This paper aims at the deficiency of information gain variable selection method(IGR-SIS)to make an improvement.The paper first proposes a variable selection method based on information gain ratio(IGR-SIS)which is suitable for the ultra-high dimensional and model-free assumption framework,then proves the deterministic selection property of IGR-SIS method theoretically,and finally verifies the effectiveness of IGR-SIS method in selecting ultra-high dimensional classified variables by Monte Carlo numerical simulation and gene expression classification data.

关 键 词：变量选择 超高维 信息增益率 确定独立筛选 

分 类 号：O21[理学—概率论与数理统计]