A New Class of IMSE-Based Criteria for Optimal Designs in Multi-response Random Coefficient Regression Models

Abstract

A new class of criteria for optimal designs in random coefficient regression (RCR) models with r responses is presented, which is based on the integrated mean squared error (IMSE) for the prediction of random effects. This class, referred to as

{\text{IMSE}}_{r,L}

\documentclass[12pt]{minimal}\usepackage{amsmath}\usepackage{wasysym}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{amsbsy}\usepackage{mathrsfs}\usepackage{upgreek}\setlength{\oddsidemargin}{-69pt}\begin{document}$$\textrm{IMSE}_{r,L}$$\end{document}

-class of criteria, is invariant with respect to different parameterizations of the model and contains

\text{IMSE}

\documentclass[12pt]{minimal}\usepackage{amsmath}\usepackage{wasysym}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{amsbsy}\usepackage{mathrsfs}\usepackage{upgreek}\setlength{\oddsidemargin}{-69pt}\begin{document}$$\textrm{IMSE}$$\end{document}

- and G-optimality as special cases for the prediction in univariate response situations. General equivalence theorems for

{\text{IMSE}}_{r,L}

\documentclass[12pt]{minimal}\usepackage{amsmath}\usepackage{wasysym}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{amsbsy}\usepackage{mathrsfs}\usepackage{upgreek}\setlength{\oddsidemargin}{-69pt}\begin{document}$$\textrm{IMSE}_{r,L}$$\end{document}

-criteria are established for

L\in [1,\infty )

\documentclass[12pt]{minimal}\usepackage{amsmath}\usepackage{wasysym}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{amsbsy}\usepackage{mathrsfs}\usepackage{upgreek}\setlength{\oddsidemargin}{-69pt}\begin{document}$$L\in [1,\infty )$$\end{document}

and

L=\infty

\documentclass[12pt]{minimal}\usepackage{amsmath}\usepackage{wasysym}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{amsbsy}\usepackage{mathrsfs}\usepackage{upgreek}\setlength{\oddsidemargin}{-69pt}\begin{document}$$L=\infty $$\end{document}

, respectively, which are used to check

{\text{IMSE}}_{r,L}

\documentclass[12pt]{minimal}\usepackage{amsmath}\usepackage{wasysym}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{amsbsy}\usepackage{mathrsfs}\usepackage{upgreek}\setlength{\oddsidemargin}{-69pt}\begin{document}$$\textrm{IMSE}_{r,L}$$\end{document}

-optimality of designs.

{\text{IMSE}}_{r,L}

\documentclass[12pt]{minimal}\usepackage{amsmath}\usepackage{wasysym}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{amsbsy}\usepackage{mathrsfs}\usepackage{upgreek}\setlength{\oddsidemargin}{-69pt}\begin{document}$$\textrm{IMSE}_{r,L}$$\end{document}

-optimal designs for linear and quadratic bi-response RCR models are given for illustration.

Keywords

Optimal designs / Integrated mean squared error matrix / IMSE-optimality / Prediction / Mixed effects model

References

[1] ChristensenRPlane Answers to Complex Questions: The Theory of Linear Models, 2002New YorkSpringer.

[2] DanskinJMThe Theory of Max–Min and Its Application to Weapons Allocation Problems, 1967New YorkSpringer.

[3] DetteH, O’BrienTE. Optimality criteria for regression models based on predicted variance. Biometrika, 1999, 86: 93-106.

[4] FedorovVV, HacklPModel-Oriented Design of Experiments, 1997New YorkSpringer.

[5] GladitzJ, PilzJ. Construction of optimal designs in random coefficient regression models. Statistics, 1982, 13: 371-385

[6] HendersonCR, KempthorneO, SearleSR, von KrosigkCM. The estimation of environmental and genetic trends from records subject to culling. Biometrics, 1959, 15: 192-218.

[7] HendersonCR. Best linear unbiased estimation prediction under a selection model. Biometrics, 1975, 31: 423-477.

[8] JensenKL, SpiildH, ToftumJ. Implementation of multivariate linear mixed-effects models in the analysis of indoor climate performance experiments. Int. J. Biometeorol., 2012, 56: 129-136.

[9] KrafftO, SchaeferM. D\documentclass[12pt]{minimal}\usepackage{amsmath}\usepackage{wasysym}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{amsbsy}\usepackage{mathrsfs}\usepackage{upgreek}\setlength{\oddsidemargin}{-69pt}\begin{document}$$D$$\end{document}-optimal designs for a multivariate regression model. J. Multivar. Anal., 1992, 42: 130-140.

[10] LairdNM, WareJH. Random-effects models for longitudinal data. Biometrics, 1982, 38: 963-974.

[11] LiuX, YueR-X, HickernellFJ. Optimality criteria for multiresponse linear models based on predictive ellipsoids. Stat. Sin., 2011, 21: 421-432

[12] LiuX, YueR-X, LinDK. Optimal design for prediction in multiresponse linear models based on rectangular confidence region. J. Stat. Plan. Inference, 2013, 143: 1954-1967.

[13] LiuX, YueR-X, WongWK. D\documentclass[12pt]{minimal}\usepackage{amsmath}\usepackage{wasysym}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{amsbsy}\usepackage{mathrsfs}\usepackage{upgreek}\setlength{\oddsidemargin}{-69pt}\begin{document}$$D$$\end{document}-optimal designs for multi-response linear mixed models. Metrika, 2019, 82: 87-98.

[14] LiuX, YueRX, ChatterjeeK. Geometric characterization of D\documentclass[12pt]{minimal}\usepackage{amsmath}\usepackage{wasysym}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{amsbsy}\usepackage{mathrsfs}\usepackage{upgreek}\setlength{\oddsidemargin}{-69pt}\begin{document}$$D$$\end{document}-optimal designs for random coefficient regression models Statist. Prob. Lett., 2020, 159. 108696

[15] MasoudiE, HollingH, WongWK. Application of imperialist competitive algorithm to find minimax and standardized maximin optimal designs. Comput. Stat. Data Anal., 2017, 113: 330-345.

[16] PilzJBayesian Estimation and Experimental Design in Linear Regression Models, 1983LeipzigTeubner

[17] PinheiroJC, BatesDMMixed Effects Models in S and S-Plus, 2000New YorkSpringer.

[18] Prus, M.: Optimal designs for the prediction in hierarchical random coefficient regression models. Ph.D. Thesis. Otto-von-Guericke University, Magdeburg (2015)

[19] PrusM, SchwabeR. Optimal designs for the prediction of individual parameters in hierarchical models. J. R. Stat. Soc. B, 2016, 78: 175-191.

[20] PrusM. Various optimality criteria for the prediction of individual response curves. Stat. Probab. Lett., 2019, 146: 36-41.

[21] PrusM. Optimal designs for minimax-criteria in random coefficient regression models. Stat. Papers, 2019, 60: 465-478.

[22] PukelsheimFOptimal Design of Experiments, 1993New YorkWiley

[23] SchmelterT. The optimality of single-group designs for certain mixed models. Metrika, 2007, 65: 183-193.

[24] SpjøtvollE. Random coefficients regression models: a review. Math. Operforsch. Stat. Ser. Stat., 1977, 8: 69-93

[25] VerbekeG, MolenberghsGLinear Mixed Models for Longitudinal Data, 2000New YorkSpringer

[26] WangCM, LamCT. A mixed-effects model for the analysis of circular measurements. Technometrics, 1997, 39: 119-126.

[27] WhittleP. Some general points in the theory of optimal experimental designs. J. R. Stat. Soc. Ser. B, 1973, 35: 123-130.

[28] WongWK. A unified approach to the construction of minimax designs. Biometrika, 1992, 79: 611-619.

[29] WongWK, CookRD. Heteroscedastic G\documentclass[12pt]{minimal}\usepackage{amsmath}\usepackage{wasysym}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{amsbsy}\usepackage{mathrsfs}\usepackage{upgreek}\setlength{\oddsidemargin}{-69pt}\begin{document}$$G$$\end{document}-optimal designs. J. R. Stat. Soc. Ser. B, 1993, 55: 871-880.

Funding

National Natural Science Foundation of China(11971318)

Natural Science Foundation of Anhui Province (CN)(2008085QA15)

Shanghai Rising-Star Program(20QA1407500)

RIGHTS & PERMISSIONS

School of Mathematical Sciences, University of Science and Technology of China and Springer-Verlag GmbH Germany, part of Springer Nature

Just Accepted

This article has successfully passed peer review and final editorial review, and will soon enter typesetting, proofreading and other publishing processes. The currently displayed version is the accepted final manuscript. The officially published version will be updated with format, DOI and citation information upon launch. We recommend that you pay attention to subsequent journal notifications and preferentially cite the officially published version. Thank you for your support and cooperation.