(PR#6452) Re: [Rd] predict.princomp fails if princomp was called

[ripley at stats.ox.ac.uk]

Yes, and it should fail or return NA, as the required information is not
available. The problem is that it had logical NAs not numeric NAs for the
center component.

#3682 points out something that is on the help page: if you supply covmat 
as a matrix (and not a covariance list) there is no information on the 
centring done and x is ignored.

On Thu, 22 Jan 2004 joehl at gmx.de wrote:

> 
> Dear all,
> 
> predict.princomp seems to expect non-NA component $center which is not set
> by princomp.default.
> 
> 
> Best regards
> 
> 
> 
> Jens Oehlschlägel
> 
> 
> 
> > x <- matrix(rnorm(400), ncol=4)
> > predict(princomp(x, covmat=cov(x)), newdata=x[1:10, ])
> Error in if (center) { : missing value where TRUE/FALSE needed
> In addition: Warning message: 
> the condition has length > 1 and only the first element will be used in: if
> (center) { 
> > version
>          _              
> platform i386-pc-mingw32
> arch     i386           
> os       mingw32        
> system   i386, mingw32  
> status                  
> major    1              
> minor    8.1            
> year     2003           
> month    11             
> day      21             
> language R
> 
> 

-- 
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