criteria.GL {MOODE} | R Documentation |
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 |
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