Help for package MOODE

Type:

Package

Title:

Multi-Objective Optimal Design of Experiments

Version:

1.0.1

Description:

Provides functionality to generate compound optimal designs for targeting the multiple experimental objectives directly, ensuring that the full set of research questions is answered as economically as possible. Designs can be found using point or coordinate exchange algorithms combining estimation, inference and lack-of-fit criteria that account for model inadequacy. Details and examples are given by Koutra et al. (2024) <doi:10.48550/arXiv.2412.17158>.

License:

GPL (≥ 3)

URL:

https://github.com/vkstats/MOODE

BugReports:

https://github.com/vkstats/MOODE/issues

Imports:

cli, far, progressr, Rdpack, rlang

Suggests:

doFuture, foreach

RdMacros:

Rdpack

Encoding:

UTF-8

NeedsCompilation:

RoxygenNote:

7.3.2

Packaged:

2025-01-08 05:43:28 UTC; vallia

Author:

Vasiliki Koutra

[aut, cre, cph], Olga Egorova [aut, cph], Steven Gilmour [aut, cph], Luzia Trinca [aut, cph]

Maintainer:

Vasiliki Koutra <vasiliki.koutra@kcl.ac.uk>

Repository:

CRAN

Date/Publication:

2025-01-08 16:10:02 UTC

MOODE: Multi-Objective Optimal Design of Experiments

Description

Author(s)

Maintainer: Vasiliki Koutra vasiliki.koutra@kcl.ac.uk (ORCID) [copyright holder]

Authors:

Olga Egorova [copyright holder]
Steven Gilmour [copyright holder]
Luzia Trinca [copyright holder]

Searching for a multi-objective optimal completely randomised design.

Description

Performing search for a (nearly) optimum factorial design, optimising with respect to a specified compound criterion.

Usage

Search(
  mood.object,
  algorithm = c("ptex", "coordex"),
  parallel = FALSE,
  verbose = TRUE
)

Arguments

mood.object

The object generated by mood function, containing the parameters of the experiment, the compound criterion and search parameters

algorithm

Parameter specifying the search algorithm. If ptex (default for K<=4), the point-exchange algorithm is used, and if coordex (default for K>4), the coordinate-exchange.

parallel

If TRUE use the doFuture package to run independent iterations of the algorithm in parallel using foreach. Requires doFuture library to be installed and a Future plan to be specified. See examples.

verbose

If TRUE, progress messages through the search iterations are shown.

Details

Search takes the mood object as an input with all the parameters of the experiment. Runs a point-exchange or a coordinate-exchange algorithm, returns design and model matrices, computation time and criteria values. See (Koutra et al. 2024) for examples of using parallel = TRUE.

Value

List of the outputs generated by the search:

X.design Design matrix.
df The number of pure error degrees of freedom.
X1 Primary model matrix for the found (nearly-) optimum design.
X2 Model matrix of potential terms for the found (nearly-) optimum design.
compound.value The compound criterion value of the (nearly-) optimum design.
criteria.values Component criteria values of the (nearly-) optimum design.
path The "path" of compound criterion values of the optimum designs obtained after for each random start.
time Computation time.
algorithm Point exchange or coordinate exchange used to find the design?
parallel Were different runs of the algorithm performed across different CPU cores (TRUE/FALSE)

References

Koutra V, Egorova O, Gilmour SG, Trinca LA (2024). “MOODE: An R Package for Multi-Objective Optimal Design of Experiments.” arXiv:2412.17158, https://arxiv.org/abs/2412.17158.

Examples


example1 <- mood(K = 2, Levels = 3, Nruns = 10, criterion.choice = "GDP", 
                kappa = list(kappa.Ds = 1./3, kappa.DP = 1./3, kappa.LoF = 1./3), 
                control = list(tau2 = 0.1), 
                model_terms = list(primary.model = "first_order", 
                potential.terms = c("x12", "x22", "x1x2")))
# Using point exchange
Search_point <- Search(example1, algorithm = 'ptex')
Search_point
# Using coordinate exchange (the default for K>4)
Search_coord <- Search(example1)
Search_coord

Forms the full candidate set of treatments for all polynomial terms

Description

This function forms the full extended candidate set with all polynomial terms, with labels, not orthonormalised.

Usage

candidate_set_full(cand.trt, K)

Arguments

cand.trt

Candidate set of treatments, the first column contains treatment labels. Usually obtained as output from the candidate_trt_set function.

K

Number of factors.

Value

The full extended candidate set: the column of treatment labels, then named columns with polynomial terms up to the 4th order. For example, "x12" stands for x_1^2, and "x1x2 stands for x_1x_2, and "x23x4" for x_2^3x_4.

Examples


# Full extended candidate set for two 3-level factors

K<-2; Levels <- rep(list(1:3),K);
candidate_set_full(candidate_trt_set(Levels, K), K)

Forms the orthonormalised full candidate set

Description

This function forms the full extended orthonormalised candidate set of primary and potential terms, with an intercept column and labels.

Usage

candidate_set_orth(cand.full, primary.terms, potential.terms)

Arguments

cand.full

Candidate set containing terms up to 4th order, with labels in the first column.

primary.terms

Character vector identifying primary model terms.

potential.terms

Character vector identifying potential model terms.

Value

The orthonormalised full candidate set containing primary and potential terms, with labels.

Examples


# Full extended orthonormalised candidate set for two 4-level factors, 
# full quadratic polynomial model as primary model and all three-order terms as potential.

K<-2; Levels <- rep(list(1:4),K)
cand.trt <- candidate_trt_set(Levels, K)
cand.full <- candidate_set_full(cand.trt, K)
prime.terms <- colnames(cand.full)[2:7]
poten.terms <- colnames(cand.full)[8:11]
Parameters <- c(1, rep(1,K), rep(1,K), K*(K-1)/2) 
candidate_set_orth(cand.full, prime.terms, poten.terms)

Forms the labelled candidate set of treatments

Description

This function forms the candidate set of treatments from the factors' levels, adds labels, with optional spherical transformation of the coordinates.

Usage

candidate_trt_set(Levels, K, Hypercube = TRUE)

Arguments

Levels

Levels of each factor.

K

Number of factors.

Hypercube

Indicates if the experimental region is a hypercube ('TRUE') or spherical ('FALSE').

Value

Matrix of candidate set of treatments, with treatment labels contained in the first column.

Examples


# Candidate treatment set for five 3-level factors

K<-5; Levels <- rep(list(1:3),K);
candidate_trt_set(Levels, K)

Calculates the values of the Generalised Ds-criterion and its components

Description

This function evaluates the Generalised Ds-criterion (Goos et al. 2005) for given primary and potential model matrices.

Usage

criteria.GD(X1, X2, search.object, eps = 1e-23)

Arguments

X1

The primary model matrix, with the first column containing the labels of treatments, and the second – the intercept term.

X2

The matrix of potential terms, with the first column containing the labels of treatments.

search.object

Object of class mood() specifying experiment parameters.

eps

Computational tolerance, the default value is 10^-23

Value

A list of values: indicator of whether the evaluation was successful ("eval"), Ds-criterion value – intercept excluded ("Ds"), Lack-of-fit criterion value ("LoF"), the bias component value ("bias"), the number of pure error degrees of freedom ("df") and the value of the compound criterion ("compound").

References

Goos P, Kobilinsky A, O'Brien TE, Vandebroek M (2005). “Model-Robust and Model-Sensitive Designs.” Computational Statistics and Data Analysis, 49, 201-216.

Examples

#Experiment: one 5-level factor, primary model -- full quadratic, one potential (cubic) term
# setting up the example
ex.mood <- mood(K = 1, Levels = 5, Nruns = 7, criterion.choice = "GDP", 
               kappa = list(kappa.Ds = 1./3, kappa.LoF = 1./3, kappa.bias = 1./3), 
               model_terms = list(primary.model = "second_order", potential.model = "cubic_terms"))
# Generating candidate set: orthonormalised
K <- ex.mood$K
Levels <- ex.mood$Levels 
cand.not.orth <- candidate_set_full(candidate_trt_set(Levels, K), K)
cand.full.orth <- candidate_set_orth(cand.not.orth, ex.mood$primary.terms, ex.mood$potential.terms)
# Choosing a design
index <- c(rep(1, 2), 3, 4, rep(5, 3))
X.primary <- cand.full.orth[index, c(1, match(ex.mood$primary.terms, colnames(cand.full.orth)))]
X.potential <- cand.full.orth[index, 
(c(1, match(ex.mood$potential.terms, colnames(cand.full.orth))))]
# Evaluating a compound GD-criterion
criteria.GD(X1 = X.primary, X2 = X.potential, ex.mood)
# Output: eval = 1, Ds = 0.7334291, LoF = 0.7212544, bias = 1.473138, df = 3, compound = 0.9202307

Calculates the values of the Generalised DPs-criterion and its components

Description

This function evaluates the Generalised DPs-criterion for given primary and potential model matrices. Components: Ds-, DPs-, LoF(DP)- and Bias(D)-optimality.

Usage

criteria.GDP(X1, X2, search.object, eps = 10^-23)

Arguments

X1

The primary model matrix, with the first column containing the labels of treatments, and the second – the intercept term.

X2

The matrix of potential terms, with the first column containing the labels of treatments.

search.object

Object of class mood() specifying experiment parameters.

eps

Computational tolerance, the default value is 10^-23

Value

A list of values: indicator of whether the evaluation was successful ("eval"), Ds-criterion value – intercept excluded ("Ds"), DPs-criterion value – intercept excluded ("DPs"), Lack-of-fit(DP) criterion value ("LoF"), the bias component value ("bias"), the number of pure error degrees of freedom ("df") and the value of the compound criterion ("compound").

Examples

# Experiment: one 5-level factor, primary model -- full quadratic, X^3 and X^4 potential terms.
ex.mood <- mood(K = 1, Levels = 5, Nruns = 8, criterion.choice = "GDP", 
               kappa = list(kappa.Ds = .25, kappa.LoF = .25, kappa.bias = .25, kappa.DP = .25), 
               model_terms = list(primary.model = "second_order", potential.terms = "x14"))
# Generating candidate sets: primary and full orthonormalised ones
K <- ex.mood$K
Levels <- ex.mood$Levels 
cand.not.orth <- candidate_set_full(candidate_trt_set(Levels, K), K)
cand.full.orth <- candidate_set_orth(cand.not.orth, ex.mood$primary.terms, ex.mood$potential.terms)
# Choosing a design
index <- c(rep(1,2),3,rep(4,2),rep(5,3))
X.primary <- cand.full.orth[index, c(1, match(ex.mood$primary.terms, colnames(cand.full.orth)))]
X.potential <- cand.full.orth[index, 
(c(1, match(ex.mood$potential.terms, colnames(cand.full.orth))))]
# Evaluating a compound GDP-criterion
criteria.GDP(X1 = X.primary, X2 = X.potential, ex.mood)

# Output: eval = 1, Ds = 0.6884783, DP = 4.4538023, LoF = 3.895182, 
# bias = 1.03807, df = 4, compound = 2.465318

Calculates the values of the Generalised L-criterion and its components

Description

This function evaluates the Generalised L-criterion (Goos et al. 2005) for given primary and potential model matrices.

Usage

criteria.GL(X1, X2, search.object, eps = 10^-23)

Arguments

X1

The primary model matrix, with the first column containing the labels of treatments, and the second – the intercept term.

X2

The matrix of potential terms, with the first column containing the labels of treatments.

search.object

Object of class mood() specifying experiment parameters.

eps

Computational tolerance, the default value is 10^-23

Value

A list of values: indicator of whether the evaluation was successful ("eval"), Ls-criterion value – intercept excluded ("Ls"), Lack-of-fit criterion value ("LoF"), the bias component value ("bias"), the number of pure error degrees of freedom ("df") and the value of the compound criterion ("compound").

References

Goos P, Kobilinsky A, O'Brien TE, Vandebroek M (2005). “Model-Robust and Model-Sensitive Designs.” Computational Statistics and Data Analysis, 49, 201-216.

Examples

#Experiment: one 5-level factor, primary model -- full quadratic, one potential (cubic) term
# setting up the example
ex.mood <- mood(K = 1, Levels = 5, Nruns = 7, criterion.choice = "GL", 
               kappa = list(kappa.L = 1./3, kappa.LoF = 1./3, kappa.bias = 1./3), 
               model_terms = list(primary.model = "second_order", potential.model = "cubic_terms"))
# Generating candidate set: orthonormalised
K <- ex.mood$K
Levels <- ex.mood$Levels 
cand.not.orth <- candidate_set_full(candidate_trt_set(Levels, K), K)
cand.full.orth <- candidate_set_orth(cand.not.orth, ex.mood$primary.terms, ex.mood$potential.terms)
# Choosing a design
index <- c(rep(1, 2), 3, 4, rep(5, 3))
X.primary <- cand.full.orth[index, c(1, match(ex.mood$primary.terms, colnames(cand.full.orth)))]
X.potential <- cand.full.orth[index, 
(c(1, match(ex.mood$potential.terms, colnames(cand.full.orth))))]
# Evaluating a compound GD-criterion
criteria.GL(X1 = X.primary, X2 = X.potential, ex.mood)
# Output: eval = 1, L = 0.3118626, LoF = 0.7212544, bias = 1.473138, df = 3, compound = 0.6919878

Calculates the values of the Generalised LP-criterion and its components

Description

This function evaluates the Generalised LP-criterion for given primary and potential model matrices. Components: L-, LP-, LoF(LP)- and Bias(L)-optimality.

Usage

criteria.GLP(X1, X2, search.object, eps = 1e-23)

Arguments

X1

The primary model matrix, with the first column containing the labels of treatments, and the second – the intercept term.

X2

The matrix of potential terms, with the first column containing the labels of treatments.

search.object

Object of class mood() specifying experiment parameters.

eps

Computational tolerance, the default value is 10^-23

Value

A list of values: indicator of whether the evaluation was successful ("eval"), Ls-criterion value – intercept excluded ("L"), LP-criterion value – intercept excluded ("LP"), Lack-of-fit(LP) criterion value ("LoF"), the bias component value ("bias"), the number of pure error degrees of freedom ("df") and the value of the compound criterion ("compound").

Examples

#' # Experiment: one 5-level factor, primary model -- full quadratic, X^3 and X^4 potential terms.
ex.mood <- mood(K = 1, Levels = 5, Nruns = 8, criterion.choice = "GLP", 
               kappa = list(kappa.L = .25, kappa.LoF = .25, kappa.bias = .25, kappa.LP = .25), 
               model_terms = list(primary.model = "second_order", potential.terms = "x14"))
# Generating candidate sets: primary and full orthonormalised ones
K <- ex.mood$K
Levels <- ex.mood$Levels 
cand.not.orth <- candidate_set_full(candidate_trt_set(Levels, K), K)
cand.full.orth <- candidate_set_orth(cand.not.orth, ex.mood$primary.terms, ex.mood$potential.terms)
# Choosing a design
index <- c(rep(1,2),3,rep(4,2),rep(5,3))
X.primary <- cand.full.orth[index, c(1, match(ex.mood$primary.terms, colnames(cand.full.orth)))]
X.potential <- cand.full.orth[index, 
(c(1, match(ex.mood$potential.terms, colnames(cand.full.orth))))]
# Evaluating a compound GDP-criterion
criteria.GLP(X1 = X.primary, X2 = X.potential, ex.mood)
# Output: eval = 1, L = 0.2952603, LP = 4.584705, LoF = 3.895182, 
# bias = 1.03807, df = 4, compound = 1.529564

Calculates the values of the MSE DPs-criterion and its components

Description

This function evaluates the MSE DPs-criterion for given primary and potential model matrices. Candidate full model matrices do not have to be orthonormalised. Components: DPs-, LoF(DP)- and MSE(D)-optimality.

Usage

criteria.mseD(X1, X2, search.object, eps = 1e-23)

Arguments

X1

The primary model matrix, with the first column containing the labels of treatments, and the second – the intercept term.

X2

The matrix of potential terms, with the first column containing the labels of treatments.

search.object

Object of class mood() specifying experiment parameters.

eps

Computational tolerance, the default value is 10^-23

Value

A list of values: indicator of whether the evaluation was successful ("eval"), DPs-criterion value – intercept excluded ("DP"), Lack-of-fit(DP) criterion value ("LoF"), the MSE(D) component value ("mse"), the number of pure error degrees of freedom ("df") and the value of the compound criterion ("compound").

Examples

# Experiment: one 5-level factor, primary model -- full quadratic, X^3 and X^4 potential terms.
set.seed(20210930)
ex.mood <- mood(K = 1, Levels = 5, Nruns = 8, criterion.choice = "MSE.D", 
               kappa = list(kappa.DP = 1./3, kappa.LoF = 1./3, kappa.mse = 1./3), 
               control = list(Biter = 1000), 
               model_terms = list(primary.model = "second_order", potential.terms = "x14"))
# Generating candidate sets: primary and full orthonormalised ones
K <- ex.mood$K
Levels <- ex.mood$Levels 
cand.not.orth <- candidate_set_full(candidate_trt_set(Levels, K), K)
cand.full.orth <- candidate_set_orth(cand.not.orth, ex.mood$primary.terms, ex.mood$potential.terms)
# Choosing a design
index <- c(rep(1,2),3,rep(4,2),rep(5,3))
X.primary <- cand.full.orth[index, c(1, match(ex.mood$primary.terms, colnames(cand.full.orth)))]
X.potential <- cand.full.orth[index, 
(c(1, match(ex.mood$potential.terms, colnames(cand.full.orth))))]
# Evaluating a compound GDP-criterion
criteria.mseD(X.primary, X.potential, ex.mood)
# Output: eval = 1, DP = 4.538023, LoF = 3.895182, mse = 0.6986903, df = 4, compound = 2.310728

Calculates the values of the MSE LP-criterion and its components

Description

This function evaluates the MSE LP-criterion for given primary and potential model matrices. Candidate full model matrices do not have to be orthonormalised. Components: LP-, LoF(LP)- and MSE(L)-optimality.

Usage

criteria.mseL(X1, X2, search.object, eps = 10^-23)

Arguments

X1

The primary model matrix, with the first column containing the labels of treatments, and the second – the intercept term.

X2

The matrix of potential terms, with the first column containing the labels of treatments.

search.object

Object of class mood() specifying experiment parameters.

eps

Computational tolerance, the default value is 10^-23

Value

A list of values: indicator of whether the evaluation was successful ("eval"), LP-criterion value – intercept excluded ("LP"), Lack-of-fit(LP) criterion value ("LoF"), the MSE(L) component value ("mse"), the number of pure error degrees of freedom ("df") and the value of the compound criterion ("compound").

Examples

#'# Experiment: one 5-level factor, primary model -- full quadratic, X^3 and X^4 potential terms.
ex.mood <- mood(K = 1, Levels = 5, Nruns = 8, criterion.choice = "MSE.L", 
               kappa = list(kappa.LP = 1./3, kappa.LoF = 1./3, kappa.mse = 1./3), 
               model_terms = list(primary.model = "second_order", potential.terms = "x14"))
# Generating candidate sets: primary and full orthonormalised ones
K <- ex.mood$K
Levels <- ex.mood$Levels 
cand.not.orth <- candidate_set_full(candidate_trt_set(Levels, K), K)
cand.full.orth <- candidate_set_orth(cand.not.orth, ex.mood$primary.terms, ex.mood$potential.terms)
# Choosing a design
index <- c(rep(1,2),3,rep(4,2),rep(5,3))
X.primary <- cand.full.orth[index, c(1, match(ex.mood$primary.terms, colnames(cand.full.orth)))]
X.potential <- cand.full.orth[index, 
(c(1, match(ex.mood$potential.terms, colnames(cand.full.orth))))]
# Evaluating a compound GDP-criterion
criteria.mseL(X.primary, X.potential, ex.mood)
# Output: eval = 1, LP = 4.584705, LoF = 3.895182, mse = 0.3926842, df = 4, compound = 1.914084

Calculates the values of the MSE DPs-criterion using the point prior for the MSE(D)-component estimation

Description

This function evaluates the MSE DPs-criterion for given primary and potential model matrices, using point MSE(D)-component estimation. Candidate full model matrices do not have to be orthonormalised. Components: DPs-, LoF(DP)- and MSE(D)-optimality.

Usage

criteria.mseP(X1, X2, search.object, eps = 10^-23)

Arguments

X1

The primary model matrix, with the first column containing the labels of treatments, and the second – the intercept term.

X2

The matrix of potential terms, with the first column containing the labels of treatments.

search.object

Object of class mood() specifying experiment parameters.

eps

Computational tolerance, the default value is 10^-23

Value

Examples

# Experiment: one 5-level factor, primary model -- full quadratic, X^3 and X^4 potential terms.
ex.mood <- mood(K = 1, Levels = 5, Nruns = 8, criterion.choice = "MSE.P", 
               kappa = list(kappa.DP = 1./3, kappa.LoF = 1./3, kappa.mse = 1./3), 
               model_terms = list(primary.model = "second_order", potential.terms = "x14"))
# Generating candidate sets: primary and full orthonormalised
K <- ex.mood$K
Levels <- ex.mood$Levels 
cand.not.orth <- candidate_set_full(candidate_trt_set(Levels, K), K)
cand.full.orth <- candidate_set_orth(cand.not.orth, ex.mood$primary.terms, ex.mood$potential.terms)
# Choosing a design
index <- c(rep(1,2),3,rep(4,2),rep(5,3))
X.primary <- cand.full.orth[index, c(1, match(ex.mood$primary.terms, colnames(cand.full.orth)))]
X.potential <- cand.full.orth[index, 
(c(1, match(ex.mood$potential.terms, colnames(cand.full.orth))))]
# Evaluating a compound GDP-criterion
criteria.mseP(X.primary, X.potential, ex.mood)
# Output: eval = 1, DP = 4.538023, LoF = 3.895182, mse = 0.6992699, df = 4, compound = 2.312135

Evaluating individual criteria of the designs, from the Generalized compound criteria (Goos et al. 2005; Egorova,~O. 2017)

Description

Calculating values of determinant- and trace-based components of Generalized D-, DP-, L- and LP- criteria for an output of a search object, with model and control parameters set in a mood object.

Usage

criteria.values.G(search.obj, mood.obj, eps = 10^-23)

Arguments

search.obj

Output of the ‘Search’ function

mood.obj

Output of the ‘mood’ function

eps

Computational tolerance, default 10^-20

Value

List of the calculated values:

df pure error degrees of freedom
Ds Ds-criterion value, intercept excluded
DP DPs-criterion value, intercept excluded
LoFD LoF(D)-criterion value from the GD-criterion
LoFDP LoF(DP)-criterion value from the GDP-criterion
biasD bias(D)-criterion value from the GD-criterion
Ls Ls-criterion value, intercept excluded
LP LPs-criterion value, intercept excluded
LoFL LoF(L)-criterion value from the GL-criterion
LoFLP LoF(LP)-criterion value from the GLP-criterion
biasL bias(L)-criterion value from the GL-criterion

References

Egorova,~O. (2017). Optimal Design of Experiments for Multiple Objectives. Ph.D. thesis, University of Southampton.

Goos P, Kobilinsky A, O'Brien TE, Vandebroek M (2005). “Model-Robust and Model-Sensitive Designs.” Computational Statistics and Data Analysis, 49, 201-216.

Evaluating individual criteria of the designs

Description

Calculating values of determinant- and trace-based components of MSE(D)- and MSE(L)- criteria for an output of a search object, with model and control parameters set in a mood object.

Usage

criteria.values.mse(search.obj, mood.obj, eps = 10^-20, Biter = 1000)

Arguments

search.obj

Output of the ‘Search’ function

mood.obj

Output of the ‘mood’ function

eps

Computational tolerance, default 10^-20

Biter

MC sample size for evaluating the mse(D)-component

Value

List of the calculated values:

df pure error degrees of freedom
Ds Ds-criterion value, intercept excluded
DP DPs-criterion value, intercept excluded
LoFDP LoF(DP)-criterion value
mseD mse(D)-criterion value, obtained via MC sampling
mseP mse(D)-criterion value, obtained using point prior
L L-criterion value, intercept excluded
LP LP-criterion value, intercept excluded
LoFLP LoF(LP)-criterion value
mseL mse(L)-criterion value

Setting up the parameters of a factorial experiment to search for multi-objective optimal completely randomised design.

Description

Creates an object containing the parameters of the experiment, compound optimality criterion with the weights and parameters of the search.

Usage

mood(
  K,
  Levels,
  Nruns,
  criterion.choice = c("GD", "GL", "GDP", "GLP", "MSE.D", "MSE.L", "MSE.P"),
  kappa = list(),
  control = list(),
  prob = list(),
  model_terms = list(primary.model = "first_order")
)

Arguments

K

Number of factors.

Levels

Either (a) a common number of levels for each factor or (b) a list of length K of the vectors containing levels of each factor.

Nruns

Number of runs of the experiment.

criterion.choice

Compound criterion to be used for the optimal design search or evaluation. Possible values are:

GL, GD for Generalised D- and L-optimality (Goos et al. 2005)
GDP and GLP for Generalised DP- and LP-optimality (Gilmour and Trinca 2012)
MSE.D, MSE.L and MSE.P for compound criteria with MSE-component: determinant-based, trace-based and determinant-based but with point estimates for the MSE(D)-component

kappa

List specifying the weights used in the compound criterion. Each named entry must be between 0 and 1.

kappa.Ds Weight of the Ds-criterion (default = 1 if criterion.choice = GD)
kappa.DP Weight of the DP-criterion (default = 1 if criterion.choice = GDP)
kappa.L Weight of the L-criterion (default = 1 if criterion.choice = GL)
kappa.LP Weight of the LP-criterion (default = 1 if criterion.choice = GLP)
kappa.LoF Weight of the lack-of-fit criterion
kappa.bias Weight of the bias criterion
kappa.mse Weight of the MSE criterion (default = 1 if 'criterion.choice = MSE.*')

control

Named list specifying control parameters for the design search.

Nstarts The number of randomly generated start designs of the search algorithm (default = 10).
Cubic Indicator of whether the experimental region is cubic (TRUE, default) or spherical (FALSE).
tau2 The variance scaling parameter for the prior distribution of the potential terms (default = 1).
Biter Number of samples for evaluating the MSE determinant-based component criterion (default = 50).
MC Indicator of whether to apply a multiple comparison (Bonferroni) correction for trace-based criteria (TRUE, default) or not (FALSE).
orth Indicator of whether to orthonormalise the potential and primary terms (⁠TRUE', default) or not (⁠FALSE').

prob

Named list specifying confidence levels for DP- (prob.DP), LP- (prob.LP) and Lack-of-fit (prob.LoF) criteria (pre-Bonferroni correction). All default to 0.95.

model_terms

A list specifying the primary (fitted) and potential (biased) models with the following named elements (see Details).

primary.model The order of the fitted polynomial model. Alternatively polynomial terms can be directly specified through the primary.terms parameter.
potential.model The order of the potential/extra polynomial terms. Alternatively can be specified through the potential.terms parameter.
primary.terms Alternative specification of the primary model via character vector of the names of the primary terms.
potential.terms Alternative specification of the potential model via character vector of the names of the potential terms.

Details

The function provides different ways of specifying the levels of the factors and the models. Although some default options are provided for, e.g., criterion.choice and ⁠kappa.*⁠ values, specification of these input parameters should be carefully chosen to reflect the aims of the experiment and available prior information.

Specifying the factors and levels

If all K factors have the same number of levels, Levels parameter is used to input that number. Otherwise, Levels is set to be a list of vectors containing the values of the factors, e.g. list(1:3, 1:2, 1:4) for 3 factors with equally spaced 3, 2 and 4 levels respectively.

Specifying the fitted model and the potential terms

There are two ways to describe the primary and potential sets of model terms via the model_terms list. Named elements primary.model and potential.model can be used to specify the fitted model and the potential terms via a string form. They are used to generate the sets of primary.terms and potential.terms which alternatively can be input directly. Possible values of primary.model are:

main_effects – main effects for all the factors (default for all criteria)
first_order – main effects and linear interactions
second_order – full second order polynomial
third_order – full second order polynomial model and all interactions of degree 3
cubic – third order polynomial model with cubic terms

The intercept is always included as a primary term.

Possible elements of the potential.model vector argument:

linear_interactions – linear interactions among the factors (default for MSE criteria)
quadratic_terms – quadratic terms for all the factors
third_order_terms – all interactions of degree 3: linear-by-linear-by-linear and quadratic-by-linear terms
cubic_terms – cubic terms for all the factors
fourth_order_terms – all interactions of degree 4, similar to third_order_terms

primary.terms and potential.terms arguments are used to specify the fitted model and the potential terms explicitly – up to the total power of 4.

Single factor powers, are coded as a string starting with with "x" and followed by the index of the factor and the power: "x32". For example, x_3^2 is coded as "x32"; "x22" stands for x_2^2, and "x4" stands for the linear term x_4.
Interaction of factors' powers are coded by merging the individual factors' powers, ordered by the factors' indexes. For example, x_2^2\times x_1 is coded as "x1x22", x_3x_12x_4 – as "x12x3x4".

Value

List of parameters of the experiment, compound criterion of choice, and primary and potential model terms.

K Number of factors.
Klev Number of levels of each factor, if all factors have the same number of levels.
Levels List of length K of the vectors containing values of the factors.
Nruns Number of runs of the experiment.
criterion.choice Compound criterion to be used for the optimal design search or evaluation.
Nstarts The number of randomly generated start designs of the search algorithm.
Biter Number of samples for evaluating the MSE determinant-based component criterion.
tau2 The variance scaling parameter for the prior distribution of the potential terms.
tau The square root of tau2
Cubic Whether the experimental region is cubic (TRUE) or spherical (FALSE).
MC Indicator of the multiple comparison (Bonferroni) correction for trace-based criteria.
⁠prob.DP, prob.LP, prob.LoF, prob.LoFL⁠ Confidence levels for the DP-, LP-, lack of fit determinant- and trace-based criteria.
⁠alpha.DP, alpha.LP, alpha.LoF, alpha.LoFL⁠ Significance levels for the DP-, LP-, lack of fit determinant- and trace-based criteria.
orth Whether the candidate sets are orthonormalised (TRUE) or not (FALSE).
Z0 Z0 matrix.
W Weight matrix for Ls criterion.
primary.terms Fitted (primary) model terms.
potential.terms Potential terms.
P The number of terms in the fitted model (including intercept).
Q The number of potential terms.
⁠kappa.Ds, kappa.DP, kappa.L, kappa.LP, kappa.LoF, kappa.bias, kappa.mse⁠ Compound criterion weights.
warning.msg Warning messages.

References

Gilmour SG, Trinca LA (2012). “Optimum Design of Experiments for Statistical Inference (with discussion).” Journal of the Royal Statistical Society C, 61, 345-401.

Goos P, Kobilinsky A, O'Brien TE, Vandebroek M (2005). “Model-Robust and Model-Sensitive Designs.” Computational Statistics and Data Analysis, 49, 201-216.

Examples


example1 <- mood(K = 5, Levels = 3, Nruns = 40, criterion.choice = "GDP", 
kappa = list(kappa.Ds = 1./3, kappa.DP = 1./3, kappa.LoF = 1./3), 
control = list(Nstarts = 50, tau2 = 0.1),
model_terms = list(primary.model = "second_order", 
potential.terms = c("x12x2", "x22x3", "x32x4", "x42x5")))
example1

example2 <- mood(K = 3, Nruns = 12, Levels = list(1:3, 1:2, 1:2), criterion.choice = "MSE.L",
kappa = list(kappa.LP = 1./2, kappa.LoF = 1./4, kappa.mse = 1./4),
control = list(Nstarts = 50, tau2 = 1), 
model_terms = list(primary.terms = "first_order",
potential.terms = c("x12", "x12x2", "x12x3")))
example2

Swapping points between the current design and candidate set

Description

Performing point-exchange algorithm, extensive swap of points procedure between the current design and candidate set.

Usage

point.swap(X1, X2, cand.full, search.object)

Arguments

X1

Current fitted (primary) model matrix

X2

Current potential terms matrix

cand.full

Full candidate matrix

search.object

Object for the search

Details

point.swap is called within the Search function

Value

A list of model matrices, criteria values and whether the search needs to continue

S3 print method

Description

Prints a summary of the mood object, including parameters that define the experiment and the (compound) criterion under which the design will be sought.

Usage

## S3 method for class 'settings'
print(x, ...)

Arguments

x

mood object

...

further arguments passed to or from other methods

Value

No return value, prints summary of object to output

MOODE: Multi-Objective Optimal Design of Experiments

Description

Author(s)

See Also

Searching for a multi-objective optimal completely randomised design.

Description

Usage

Arguments

Details

Value

References

See Also

Examples

Forms the full candidate set of treatments for all polynomial terms

Description

Usage

Arguments

Value

Examples

Forms the orthonormalised full candidate set

Description

Usage

Arguments

Value

Examples

Forms the labelled candidate set of treatments

Description

Usage

Arguments

Value

Examples

Calculates the values of the Generalised Ds-criterion and its components

Description

Usage

Arguments

Value

References

Examples

Calculates the values of the Generalised DPs-criterion and its components

Description

Usage

Arguments

Value

Examples

Calculates the values of the Generalised L-criterion and its components

Description

Usage

Arguments

Value

References

Examples

Calculates the values of the Generalised LP-criterion and its components

Description

Usage

Arguments

Value

Examples

Calculates the values of the MSE DPs-criterion and its components

Description

Usage

Arguments

Value

Examples

Calculates the values of the MSE LP-criterion and its components

Description

Usage

Arguments

Value

Examples

Calculates the values of the MSE DPs-criterion using the point prior for the MSE(D)-component estimation

Description

Usage

Arguments

Value

Examples

Evaluating individual criteria of the designs, from the Generalized compound criteria (Goos et al. 2005; Egorova,~O. 2017)

Description

Usage

Arguments

Value