Robust canonical correlations: A comparative study

被引：38

作者：

Branco, JA

Croux, C

Filzmoser, P

Oliveira, MR

机构：

[1] Inst Super Tecn, Dept Math, P-1049001 Lisbon, Portugal

[2] Inst Super Tecn, Ctr Math & Applications, P-1049001 Lisbon, Portugal

[3] Katholieke Univ Leuven, Dept Appl Econ, B-3000 Louvain, Belgium

[4] Vienna Univ Technol, Dept Stat & Probabil Theory, A-1040 Vienna, Austria

来源：

COMPUTATIONAL STATISTICS | 2005年 / 20卷 / 02期

关键词：

Alternating Regressions; Canonical Correlation; Correlation Measures; Projection Pursuit; Robustness; Robust Covariance Estimation; Robust Regression;

D O I：

10.1007/BF02789700

中图分类号：

O21 [概率论与数理统计]; C8 [统计学];

学科分类号：

020208 ; 070103 ; 0714 ;

摘要：

Several approaches for robust canonical correlation analysis will be presented and discussed. A first method is based on the definition of canonical correlation analysis as looking for linear combinations of two sets of variables having maximal (robust) correlation. A second method is based on alternating robust regressions. These methods axe discussed in detail and compared with the more traditional approach to robust canonical correlation via covariance matrix estimates. A simulation study compares the performance of the different estimators under several kinds of sampling schemes. Robustness is studied as well by breakdown plots.

引用

页码：203 / 229

页数：27

共 24 条

[1]

[Anonymous], 1999, APPL MULTIVARIATE AN

[2] The largest nonidentifiable outlier: a comparison of multivariate simultaneous outlier identification rules [J].