withinpca function

Normed within principal component analysis

Performs a normed within Principal Component Analysis.

withinpca(df, fac, scaling = c("partial", "total"), 
    scannf = TRUE, nf = 2)

Arguments

• df: a data frame with quantitative variables

• fac: a factor partitioning the rows of df in classes

• scaling: a string of characters as a scaling option :

  if "partial", the sub-table corresponding to each class is centred and normed.

  If "total", the sub-table corresponding to each class is centred and the total table is then normed.

• scannf: a logical value indicating whether the eigenvalues bar plot should be displayed

• nf: if scannf FALSE, an integer indicating the number of kept axes

Details

This functions implements the 'Bouroche' standardization. In a first step, the original variables are standardized (centred and normed). Then, a second transformation is applied according to the value of the scaling

argument. For "partial", variables are standardized in each sub-table (corresponding to each level of the factor). Hence, variables have null mean and unit variance in each sub-table. For "total", variables are centred in each sub-table and then normed globally. Hence, variables have a null mean in each sub-table and a global variance equal to one.

Returns

returns a list of the sub-class within of class dudi. See wca

References

Bouroche, J. M. (1975) Analyse des données ternaires: la double analyse en composantes principales. Thèse de 3ème cycle, Université de Paris VI.

Author(s)

Daniel Chessel

Anne-Béatrice Dufour anne-beatrice.dufour@univ-lyon1.fr

Examples

data(meaudret)
wit1 <- withinpca(meaudret$env, meaudret$design$season, scannf = FALSE, scaling = "partial")
kta1 <- ktab.within(wit1, colnames = rep(c("S1", "S2", "S3", "S4", "S5"), 4))
unclass(kta1)

# See pta
plot(wit1)