Error Covariance Penalized Regression: A novel multivariate model combining penalized regression with multivariate error structure

被引：14

作者：

Allegrini, Franco ^{[1
]}

Braga, Jez W. B. ^{[1
,2
]}

Moreira, Alessandro C. O. ^{[2
,3
]}

Olivieri, Alejandro C. ^{[1
]}

机构：

[1] Univ Nacl Rosario, Inst Quim Rosario IQUIR CONICET, Fac Ciencias Bioquim & Farmaceut, Dept Quim Analit, Suipacha 531,S2002LRK, Rosario, Santa Fe, Argentina

[2] Univ Brasilia, Inst Quim, Lab Automacao Quimiometria & Quim Ambiental, BR-70904970 Brasilia, DF, Brazil

[3] Serv Florestal Brasileiro, Lab Prod Florestais, BR-70818900 Brasilia, DF, Brazil

来源：

ANALYTICA CHIMICA ACTA | 2018年 / 1011卷

关键词：

Multivariate calibration; Penalized regression; Error covariance matrix; PRINCIPAL COMPONENT ANALYSIS; CALIBRATION MAINTENANCE; TIKHONOV REGULARIZATION; SPARSE DECONVOLUTION; RIDGE-REGRESSION; SPECTROSCOPY; PARAMETER; VARIANTS; FIGURES; PENALTY;

D O I：

10.1016/j.aca.2018.02.002

中图分类号：

O65 [分析化学];

学科分类号：

070302 ; 081704 ;

摘要：

A new multivariate regression model, named Error Covariance Penalized Regression (ECPR) is presented. Following a penalized regression strategy, the proposed model incorporates information about the measurement error structure of the system, using the error covariance matrix (ECM) as a penalization term. Results are reported from both simulations and experimental data based on replicate mid and near infrared (MIR and NIR) spectral measurements. The results for ECPR are better under non-iid conditions when compared with traditional first-order multivariate methods such as ridge regression (RR), principal component regression (PCR) and partial least-squares regression (PLS). (C) 2018 Elsevier B.V. All rights reserved.

引用

页码：20 / 27

页数：8

共 42 条

[1] Multi-way figures of merit in the presence of heteroscedastic and correlated instrumental noise: Unfolded partial least-squares with residual multi-linearization [J].

Allegrini, Franco ;

Olivieri, Alejandro C. .

CHEMOMETRICS AND INTELLIGENT LABORATORY SYSTEMS, 2016, 158 :200-209

[2] Generalized error-dependent prediction uncertainty in multivariate calibration [J].

Allegrini, Franco ;

Wentzell, Peter D. ;

Olivieri, Alejandro C. .

ANALYTICA CHIMICA ACTA, 2016, 903 :51-60

[3] Applications of maximum likelihood principal component analysis: incomplete data sets and calibration transfer [J].

Andrews, DT ;

Wentzell, PD .

ANALYTICA CHIMICA ACTA, 1997, 350 (03) :341-352

[4]

[Anonymous], 1963, Soviet Math

[5] Chemometrics description of measurement error structure: Study of an ultrafast absorption spectroscopy experiment [J].

Blanchet, Lionel ;

Rehault, Julien ;

Ruckebusch, Cyril ;

Huvenne, Jean Pierre ;

Tauler, Roma ;

de Juan, Anna .

ANALYTICA CHIMICA ACTA, 2009, 642 (1-2) :19-26

[6] Figures of merit for the determination of the polymorphic purity of carbamazepone by infared spectroscopy and multivariate calibration [J].

Braga, JWB ;

Poppi, RJ .

JOURNAL OF PHARMACEUTICAL SCIENCES, 2004, 93 (08) :2124-2134

[7] Discordance between net analyte signal theory and practical multivariate calibration [J].

Brown, CD .

ANALYTICAL CHEMISTRY, 2004, 76 (15) :4364-4373

[8] Derivative preprocessing and optimal corrections for baseline drift in multivariate calibration [J].

Brown, CD ;

Vega-Montoto, L ;

Wentzell, PD .

APPLIED SPECTROSCOPY, 2000, 54 (07) :1055-1068

[9]

Brown CD, 1999, J CHEMOMETR, V13, P133, DOI 10.1002/(SICI)1099-128X(199903/04)13:2<133::AID-CEM533>3.3.CO

[10]

2-3

← 1 2 3 4 5 →