函数性广义线性模型曲线选择的正则化方法

统计研究 ›› 2012, Vol. 29 ›› Issue (9): 95-102.

函数性广义线性模型曲线选择的正则化方法

张景肖刘燕平

出版日期:2012-09-15 发布日期:2012-08-28

Regularized Methods for Curve Selection in Functional Generalized Linear Model

Zhang Jingxiao & Liu Yanping

Online:2012-09-15 Published:2012-08-28

摘要/Abstract

摘要： 本文对函数性广义线性模型曲线选择的正则化方法进行了较全面地综述，并比较了各种方法的性质。结果发现，函数性广义线性模型曲线选择问题具有群组效应，另外可能具有高维数据性质。同时通过数据模拟发现，Group Bridge、Group MCP、Elastic Net和Mnet表现出较好的数值结果。

关键词: 函数性广义线性模型, 曲线选择, 正则化方法

Abstract: This paper generally summarized the plentiful regularized methods for curve selection in functional generalized linear model. The results showed that curve selection of functional generalized linear model demonstrates group effect and also probably involves a high-dimensional feature. Finally, through data simulation, this paper found that Group Bridge, Group MCP, Elastic Net and Mnet showed good numerical results.

Key words: Functional Generalized Linear Model, Curve Selection, Regularized Methods

张景肖刘燕平. 函数性广义线性模型曲线选择的正则化方法[J]. 统计研究, 2012, 29(9): 95-102.

Zhang Jingxiao & Liu Yanping. Regularized Methods for Curve Selection in Functional Generalized Linear Model[J]. , 2012, 29(9): 95-102.