基于正则化回归的变量选择方法在高维数据中的应用  

Application of variable selection method based on regularized regression to high dimensional data

作　　者：荣雯雯 张奇[1] 刘艳[1] RONG Wen-wen, ZHANG Qi, LIU Yan(Department of Health Statistics, Harbin Medical University, Harbin, Heilongjiang 150081, Chin)

机构地区：[1]哈尔滨医科大学卫生统计学教研室,黑龙江哈尔滨150081

出　　处：《实用预防医学》2018年第6期645-648,共4页Practical Preventive Medicine

基　　金：国家自然基金(81172741;30972537)

摘　　要：变量筛选和模型估计一直是高维数据的研究热点,而高维数据的维度灾难问题日渐突出,传统的统计分析方法因模型不稳定不再适用,本文对高维数据中基于正则化回归的变量选择方法的原理、适用的数据类型及优缺点、调整参数的选择进行综述。Variable filtering and model estimation have been the hotspot of high dimensional data,and the dimensionality problem of high dimensional data is becoming more and more prominent. The traditional statistical analysis method is no longer applicable due to the instability of the model. In this paper,we review the principle of variable selection method based on regularized regression in high dimensional data,the data type and the advantages and disadvantages,and the selection of adjustment parameters.

关 键 词：高维数据 正则化 惩罚项 调整参数 

分 类 号：R195.1[医药卫生—卫生统计学]