hi,
I’m sorry, but it’s difficult question ... I haven't answer ... ;"(
To use MCOA, we need only 1 common dimension. The individual tables can
associate by the row (with the same weight of rows) or by the columns (with the
same weight of colunms). Normally, if all tables have the same dimensions (same
rows and same columns), it is not a problem.
# example 1: same rows, different columns
require(ade4)
data(escopage)
w <- data.frame(scale(escopage$tab))
w <- ktab.data.frame(w, escopage$blo, tabnames = escopage$tab.names)
for(i in 1 : length(escopage$blo))
print(dim(w[[i]]))
mcoa1 <- mcoa(w,option="uniform",scannf=F)
mcoa2 <- mcoa(w,option="lambda1", scannf=F)
mcoa3 <- mcoa(w,option="inertia", scannf=F)
mcoa1
mcoa2
mcoa3
# example 2: same rows, same columns
data(meaudret)
mil <- meaudret$mil
indic <- NULL
for(i in 1:4)
indic <- c(indic, rep(i,5))
ldudi <- list()
for(i in 1:4)
ldudi[[i]] <- dudi.pca(mil[indic==i,],scale=T,center=T,scannf=F)
ktab1 <- ktab.list.dudi(ldudi, tabnames=c("sp","su","au","wi"))
for(i in 1 : length(ktab1$blo))
print(dim(ktab1[[i]]))
mcoa1 <- mcoa(ktab1,option="uniform",scannf=F)
mcoa2 <- mcoa(ktab1,option="lambda1", scannf=F)
mcoa3 <- mcoa(ktab1,option="inertia", scannf=F)
mcoa1
mcoa2
mcoa3
# La.svd : Compute the singular-value decomposition of a rectangular matrix.
?La.svd
The problem is associated with the function La.svd and is maybe inherent in your
data set.
hope this help,
P.BADY
PS: Your French is better than my English :)
En réponse à Aedin <Aedin.Culhane@ucd.ie>:
> Bonjour,
>
> je voudrais poser une autre question sur MCOA s'il vous plait,
>
> If I run mcoa on 3 datasets with equal nrow but unequal ncol, the
> options "lambda1" and "uniform" are ok but option= "inertia" fails.
> Should I expect this?
>
> mcoa(testdata$ktab, option="lambda1")
> mcoa(testdata$ktab, option="uniform")
> mcoa(testdata$ktab, option="inertia")
> Error in La.svd(x, nu, nv) : error code 1 from Lapack routine
> dgesdd
>
> If I run the analysis on a dataset with equal nrow, ncol. I don't get
> the error with option="inertia".
>
> Is this expected?
>
> Merci
> Aedin
>
>
> (mmh.. Not so sure my french is improving!!! Any tips for me when
> talking to the many French supporters in Dublin at the weekend???)
>
>
>
---------------------------------------------------------
Pierre BADY <°)))><
Université Claude Bernard Lyon 1
UMR CNRS 5023, LEHF
bat Alphonse Forel
43 boulevard du 11 novembre 1918
F-69622 VILLEURBANNE CEDEX
FRANCE
TEL : +33 (0)4 72 44 62 34
FAX : +33 (0)4 72 43 28 92
MEL : pierre.bady@univ-lyon1.fr
http://limnologie.univ-lyon1.fr
http://pierre.bady.free.fr/ (in construction)
This archive was generated by hypermail 2b30 : Tue Oct 18 2005 - 10:35:39 MEST