Estimating the covariance matrix of the coefficient estimator in multivariate partial least squares regression with chemical applications

被引：5

作者：

Martinez, Jose L. ^{[1
]}

Leiva, Victor ^{[2
]}

Saulo, Helton ^{[3
]}

Liu, Shuangzhe ^{[4
]}

机构：

[1] Univ Sinu, Dept Basic Sci, Monteria, Colombia

[2] Pontificia Univ Catolica Valparaiso, Sch Ind Engn, Valparaiso, Chile

[3] Univ Brasilia, Dept Stat, Brasilia, DF, Brazil

[4] Univ Canberra, Fac Sci & Technol, Canberra, ACT, Australia

来源：

CHEMOMETRICS AND INTELLIGENT LABORATORY SYSTEMS | 2021年 / 214卷

关键词：

Covariance matrix; Jackknife method; Monte Carlo method; PLS regression; R software; Standard error; PREDICTION INTERVALS; MODEL; PLS;

D O I：

10.1016/j.chemolab.2021.104328

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 ;

摘要：

The partial least squares (PLS) regression is a statistical learning technique that solves collinearity and/or highdimensionality in the space of covariates. In this paper, we propose a new estimator for the covariance matrix of the estimator of the regression coefficients in the multivariate PLS model. This new estimator is simple to be calculated and with a low computational cost. We conduct a Monte Carlo simulation study to assess the performance of the proposed estimator. Then, we apply our proposal to analyze a multivariate real chemical data set. These numerical results show the excellent performance of our proposal.

引用

页数：8

共 34 条

[1]

[Anonymous], 1975, Perspectives in probability and statistics

[2] PLS generalised linear regression [J].