Lasso及其相关方法在广义线性模型模型选择中的应用

被引:0
作者
龚建朝
机构
[1] 中南大学
关键词
套索; 平滑削边绝对偏离罚函数; 自适应套索; 弹性网; 放松的套索;
D O I
暂无
年度学位
2008
学位类型
硕士
导师
摘要
模型选择在统计建模过程中是极其重要的问题,但传统的逐步回归结合信息准则的方法却存在一些不足。Tibshirani,R.(1996)针对这一问题提出了一种称之为Lasso的新的模型选择方法。该方法很好的克服了传统方法的一些不足。Efron等(2004)提出了能很好地解决Lasso问题的算法。但Lasso方法在理论上也存在一些不足,因此很多学者对此提出了改进,如:SCAD(Fan 2001)、Adaptive-lasso(Zou 2006和Wang 2007)、elastic net(Zou和Hastie 2005)、Relaxed Lasso(Nicolai Meinshausen 2007)等。 本文的工作主要有三点: (1)基于线性模型对Lasso、SCAD、Adaptive-lasso、elasticnet、Relaxed Lasso等方法进行了系统比较研究,指出了它们之间的联系; (2)基于Logistic回归通过实际数据研究了Lasso及其相关方法的一些差异,并在此基础上系统介绍了基于Lasso方法的广义线性模型模型选择的统一处理框架; (3)指出了有待进一步研究的问题。
引用
收藏
页数:48
共 30 条
[1]
Variable selection in semiparametric regression modeling [J].
Li, Runze ;
Liang, Hua .
ANNALS OF STATISTICS, 2008, 36 (01) :261-286
[2]
The Dantzig Selector: Statistical Estimation When p Is Much Larger than n.[J].Emmanuel Candes;Terence Tao.The Annals of Statistics.2007, 6
[3]
PATHWISE COORDINATE OPTIMIZATION [J].
Friedman, Jerome ;
Hastie, Trevor ;
Hoefling, Holger ;
Tibshirani, Robert .
ANNALS OF APPLIED STATISTICS, 2007, 1 (02) :302-332
[4]
On the "degrees of freedom" of the lasso [J].
Zou, Hui ;
Hastie, Trevor ;
Tibshirani, Robert .
ANNALS OF STATISTICS, 2007, 35 (05) :2173-2192
[5]
Unified LASSO estimation by least squares approximation [J].
Wang, Hansheng ;
Leng, Chenlei .
JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION, 2007, 102 (479) :1039-1048
[6]
Robust regression shrinkage and consistent variable selection through the LAD-lasso [J].
Wang, Hansheng ;
Li, Guodong ;
Jiang, Guohua .
JOURNAL OF BUSINESS & ECONOMIC STATISTICS, 2007, 25 (03) :347-355
[7]
Relaxed lasso [J].
Meinshausen, Nicolai .
COMPUTATIONAL STATISTICS & DATA ANALYSIS, 2007, 52 (01) :374-393
[8]
Adaptive lasso for Cox's proportional hazards model [J].
Zhang, Hao Helen ;
Lu, Wenbin .
BIOMETRIKA, 2007, 94 (03) :691-703
[9]
Piecewise linear regularized solution paths [J].
Rosset, Saharon ;
Zhu, Ji .
ANNALS OF STATISTICS, 2007, 35 (03) :1012-1030
[10]
Dimension reduction and coefficient estimation in multivariate linear regression.[J].MingYuan;AliEkici;ZhaosongLu;RenatoMonteiro.Journal of the Royal Statistical Society: Series B (Statistical Methodology).2007, 3