blocksdesign (version 3.6)

fullrankModel: Full rank model

Description

Finds a full column rank model for a treatment formula and a treatmnent data frame

Usage

fullrankModel(TF, model_formula)

Arguments

TF

is the treatments factor data frame

model_formula

is the matrix.model formula

Value

Maximal full rank model matrix for fitted model

Details

The treatment formula fits a model.matrix for a set of treatments which may be factorial or numeric or a mixture of both. The function will fit the specified model and if there are column dependencies the function will then reduce the column space to full rank using the QR decomposition

Examples

# NOT RUN {
# Treatments are two 2-level factors A and B and one 3-level factor V
# Required model is A + B + A:B + A:linear(V) + B:linear(V) + quadratic(V)
# The example shows model formula which appear 'correct' but which over-parameterize
# the model. It is 'reasonable' to expect model.matrix to give a full rank model and
# the QR method will ensure that the fitted model is indeed a full rank model. 
# NB The user MUST THEN CHECK to ensure that the fitted model is the required model.  

treatments = expand.grid(A = factor(1:2), B = factor(1:2), V = 1:3)

model = " ~ A * B + poly(V,2) + A:poly(V,1)  + B:poly(V,1)"
model.matrix(as.formula(model),treatments)
fullrankModel(treatments,model)

model = " ~ (A + B) * poly(V,1) + poly(V,2)  + A:B"
model.matrix(as.formula(model),treatments)
fullrankModel(treatments,model)

# }