摘要
模型选择是统计学的热点研究问题。近年来随着数据维数越来越高,传统模型选择方法的应用受到了很多制约。本文着重介绍高维模型选择的新方法,并讨论实现模型选择过程的一个重要环节,即调整参数的选取。最后文章总结归纳了未来可能的研究方向。
Model selection is an important issue in Statistics. Traditional model-selection methods, however, meet difficulties with the increasing of data dimension. This paper is devoted to the survey of the model selection methods for high-dimensional data. The choice of tuning parameters is also discussed. Some future research directions are provided.
出处
《数理统计与管理》
CSSCI
北大核心
2012年第4期640-658,共19页
Journal of Applied Statistics and Management
基金
国家自然科学基金(70625004,11021161,70933003)的资助
关键词
高维数据
模型选择
惩罚因子
降维
调整参数
high dimension, model selection, penalized factor, dimension reduction, tuning parameter.
