PCALMEffects: PCA on the effect matrices
In FranceschiniS/LMWiRe: R Routines for Linear Model with Wide Response

Description Usage Arguments Details Value Examples

Run a Principal Component Analysis on the effect matrices and adapt the result according to the method.

1	PCALMEffects(ResLMEffectMatrices, method = c("ASCA", "APCA", "ASCA-E"))

`ResLMEffectMatrices`	a ResLMEffectMatrices list from `LMEffectMatrices`
`method`	The method use to compute the PCA. One of `c("ASCA","APCA","ASCA-E")`

The function allows 3 different methods :

ASCA: The PCA is applied directly on the pure effect matrix
ASCA-E: The PCA is applied directly on the pure effect matrix but scores are updated
APCA: The PCA is applied on the augmented effect matrix

The ASCA-E method add the residual to the scores. APCA applied add the residuals to the effect matrix before the PCA.

A list of PCA results from SVDforPCA for each effect matrix. Those results contain :

scores: Scores from the PCA for each of the n components
loadings: Loadings from the PCA for each of the n component
eigval: Eigenvalues of each of the n component
pcd: Singular values of each of the n component
pcu: nxn matrix of normalized scores
var: Explained variance of each of the n component
cumvar: Cumulated explained variance of each of the n component
original.dataset: Original dataset

There are also others outputs :

method
Type3Residuals
variationPercentages

data('UCH')
ResLMModelMatrix = LMModelMatrix(formula=as.formula(UCH$formula),design=UCH$design)
ResLMEffectMatrices = LMEffectMatrices(ResLMModelMatrix,outcomes=UCH$outcomes)
ResPCALMEffects = PCALMEffects(ResLMEffectMatrices,method="ASCA-E")
 PlotScoresXY(ResPCALMEffects,UCH$design,EffectVector=c("Hippurate"),
               varname.color=c("Citrate"),varname.pch=c("Time"))