R help - Apr 2008 - significant variables in GPLS ?

Hello,
I am using the gpls package for modelling vegetation classes.

My problem is that I now want to know which input variables are significant for
the modelling of the classes to recalculate the equation again with just the
selected variables.
I think I can analyse the significance of the variables via their weights.

I used the "gpls1a" term for two group classification. Here my code:

----------------------------------------------------------------------
library(gpls)   #
> spex_Y<-read.csv("F:/GPLS/spex_Y.csv", header=TRUE,
sep=";")#, row.names="ID")  #
> spex_X<-read.csv("F:/GPLS/spex_X.csv", header=TRUE,
sep=";")#, row.names="ID")  #
>
> test <- glpls1a(spex_X, spex_Y$A_mell,K.prov=7, br=FALSE)
> names(test)
[1] "coefficients"   "convergence"    "niter"     
"family"

[5] "link"           "levs"          
"bias.reduction"


coefficients = regression coefficients

convergence = whether convergence is achieved

niter total = number of iterations

bias.reduction =  whether Firth's procedure is used

link = link function, logit is the only one practically implemented now


---------------------------------------------------------------------

But the values (coefficients, convergence, niter, family, link, levs,       
bias.reduction) I have got from the gpls1a do not contain any information about
the significance of my input variables.

Does anybody have an idea how I can get information about the significance of my
input values?

This would really help me a lot.


-- 

Jetzt dabei sein: http://www.shortview.de/?mc=sv_ext_mf at gmx

There is little theory about significance and testing for PLSR (and, I
would guess, GPLSR).  Many practicioners use Jackknife variance
estimates as a basis for significance tests.  Note, however, that these
variance estimates are known to be biased (in general), and their
distribution is (to my knowledge) not known.  Any significance deduced
from them should therefore be regarede as merely indicators.

-- 
Bj?rn-Helge Mevik