ade4 (version 1.7-19)

bca: Between-Class Analysis

Description

Performs a particular case of a Principal Component Analysis with respect to Instrumental Variables (pcaiv), in which there is only a single factor as explanatory variable.

Usage

# S3 method for dudi
bca(x, fac, scannf = TRUE, nf = 2, ...)

Value

Returns a list of class dudi, subclass 'between' containing

tab

a data frame class-variables containing the means per class for each variable

cw

a numeric vector of the column weigths

lw

a numeric vector of the class weigths

eig

a numeric vector with all the eigenvalues

rank

the rank of the analysis

nf

an integer value indicating the number of kept axes

c1

a data frame with the column normed scores

l1

a data frame with the class normed scores

co

a data frame with the column coordinates

li

a data frame with the class coordinates

call

the matching call

ratio

the bewteen-class inertia percentage

ls

a data frame with the row coordinates

as

a data frame containing the projection of inertia axes onto between axes

Arguments

x

a duality diagram, object of class dudi from one of the functions dudi.coa, dudi.pca,...

fac

a factor partitioning the rows of dudi$tab in classes

scannf

a logical value indicating whether the eigenvalues barplot should be displayed

nf

if scannf FALSE, a numeric value indicating the number of kept axes

...

further arguments passed to or from other methods

Author

Daniel Chessel
Anne-Béatrice Dufour

References

Dolédec, S. and Chessel, D. (1987) Rythmes saisonniers et composantes stationnelles en milieu aquatique I- Description d'un plan d'observations complet par projection de variables. Acta Oecologica, Oecologia Generalis, 8, 3, 403--426.

Examples

data(meaudret)
pca1 <- dudi.pca(meaudret$env, scan = FALSE, nf = 4)
pca2 <- dudi.pca(meaudret$spe, scal = FALSE, scan = FALSE, nf = 4)
bet1 <- bca(pca1, meaudret$design$site, scan = FALSE, nf = 2)
bet2 <- bca(pca2, meaudret$design$site, scan = FALSE, nf = 2)
  
if(adegraphicsLoaded()) {
  g1 <- s.class(pca1$li, meaudret$design$site, psub.text = "Principal Component Analysis (env)",
    plot = FALSE)
  g2 <- s.class(pca2$li, meaudret$design$site, psub.text = "Principal Component Analysis (spe)",
    plot = FALSE)
  g3 <- s.class(bet1$ls, meaudret$design$site, psub.text = "Between sites PCA (env)", plot = FALSE)
  g4 <- s.class(bet2$ls, meaudret$design$site, psub.text = "Between sites PCA (spe)", plot = FALSE)
  G <- ADEgS(list(g1, g2, g3, g4), layout = c(2, 2))
  
} else {
  par(mfrow = c(2, 2))
  s.class(pca1$li, meaudret$design$site, sub = "Principal Component Analysis (env)", csub = 1.75)
  s.class(pca2$li, meaudret$design$site, sub = "Principal Component Analysis (spe)", csub = 1.75)
  s.class(bet1$ls, meaudret$design$site, sub = "Between sites PCA (env)", csub = 1.75)
  s.class(bet2$ls, meaudret$design$site, sub = "Between sites PCA (spe)", csub = 1.75)
  par(mfrow = c(1, 1))
}

coib <- coinertia(bet1, bet2, scann = FALSE)
plot(coib)