[R] Built-In Wilks Lambda for lda?

Till Baumgaertel till.baumgaertel at epost.de
Mon Mar 17 02:48:49 CET 2003


using the lda-method from MASS-package I was wondering whether there is
 a built-in method for figuring out Wilks' Lambda?

Searching the Web I found in the r-help archive a thread form june 2002,
but it didn't help me. 
I understand I can use manova and its summary-method to get Wilks' Lambda
on the screen, but I don't see the connection to lda from MASS.

Or does 
  summary( manova( df[ train] ~ classes[ train]),test="Wilks")
produce Wilks Lambda as used in lda/MASS? 

In this case I would not know if an access like this
  lambda <- summary( manova( as.matrix(dataframe) ~ classes),test="Wilks")$stats[
would give me Wilk's Lambda under all circumstances.

I played a little with some data and found out this is not working every
time. manova errors sometimes "residuals have rank x < y" where "y" equals
ncol(df). But this shouln't impact Wilk's Lambda?

Thanks for your help,

Mehr Power für Ihre eMail - mit den neuen Leistungspaketen bei http://www.epost.de

More information about the R-help mailing list