摘要
变量选择是统计建模中重要的问题。当试验数据维数很高时,传统变量选择方法的应用受到了很多制约。本文以高维混料试验为基础,比较了AIC准则和LASSO在变量选择问题上的优良性。通过实例验证,LASSO可以快速且准确地对高维混料模型中的变量进行筛选,从而得出最优模型,达到降低成本、提高利益的目的。
Variable selection is an important issue in Statistical Modeling. Traditional variable selection method is difficult in dealing with the increasing dimension of experimental data. In this paper, based on the high-dimensional mixture experimental design, the optimality of the AIC criterion and LASSO on variable selection are compared. As showed by examples, LASSO can quickly and accurately filter the variables for the high-dimensional mixture model selection process. The optimal model got by the process are powerful in reducing cost and increasing benefit.
作者
冷薇
李俊鹏
张崇岐
LENG Wei;LI Jun-peng;ZHANG Chong-qi(School of Economics and Statistics Guangzhou University,Guangdong Guangzhou 510006,China)
出处
《数理统计与管理》
CSSCI
北大核心
2019年第1期81-86,共6页
Journal of Applied Statistics and Management
基金
国家自然科学基金(11671104)
广州大学研究生"基础创新"项目(2017GDJC-M49)
关键词
高维混料模型
变量选择
LASSO
high-dimensional mixture model
variable selection
LASSO
