body of non-visible function

getS3method("princomp", "default").
getAnywhere("princomp.default")

The first of these is in the See Also of ?methods.

Hello,
Is there any possibility in R to see the body of the 

non-visible

function, for
example 

princomp
?

If I do :

methods(princomp)

so, I get that princomp.default and princomp.formula are non-visible 
functions and
body(princomp.default) doesnt show it.

In particular, I guess I have a very na?ve question&
Id like to see how scores calculation is implemented in the function
princomp. Because when I multiply my data matrix on the matrix of loadings

data.matrix %*% princomp(data.matrix, scores=T)$loadings

I get different result than just doing

princomp(data.matrix, scores=T)$scores.

You have forgotten to centre your data.  It may be more helpful to look at 
the predict method.

Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

Anna Oganyan wrote on 5/5/2005 7:42 AM:

Hello,
Is there any possibility in R to see the body of the “non-visible” 
function, for
example “princomp”?
If I do :

 > methods(princomp)

so, I get that princomp.default and princomp.formula are non-visible 
functions and
body(princomp.default) doesn’t show it.

In particular, I guess I have a very naïve question…
I’d like to see how scores calculation is implemented in the function
princomp. Because when I multiply my data matrix on the matrix of loadings

 >data.matrix %*% princomp(data.matrix, scores=T)$loadings

I get different result than just doing

 >princomp(data.matrix, scores=T)$scores.

Thanks.
Anna

Hi Anna,

Use getAnywhere("princomp.default") or stats:::princomp.default. This 
has to do with princomp.default not being exported in the stats package 
NAMESPACE.

As for the difference, the scores are based on centering the columns of 
data.matrix before determining the principal components. For example:

 > X <- as.matrix(USArrests)
 > Xc <- sweep(X, 2, colMeans(X), "-")
 > pc <- princomp(X, scores = TRUE) # inappropriate, see ?princomp
 > str(pc$scores)
  num [1:50, 1:4]  -64.8  -92.8 -124.1  -18.3 -107.4 ...
  - attr(*, "dimnames")=List of 2
   ..$ : chr [1:50] "Alabama" "Alaska" "Arizona" "Arkansas" ...
   ..$ : chr [1:4] "Comp.1" "Comp.2" "Comp.3" "Comp.4"
 > Xl <- Xc %*% loadings(pc)
 > str(Xl)
  num [1:50, 1:4]  -64.8  -92.8 -124.1  -18.3 -107.4 ...
  - attr(*, "dimnames")=List of 2
   ..$ : chr [1:50] "Alabama" "Alaska" "Arizona" "Arkansas" ...
   ..$ : chr [1:4] "Comp.1" "Comp.2" "Comp.3" "Comp.4"
 > sum(abs(Xl - pc$scores))
[1] 4.97999e-12
 >

Finally, "data.matrix" is a function in the base package. I would avoid 
using it as a variable name.

--sundar