R devel - Aug 2003 - covmat argument in princomp() (PR#3682)

R version: 1.7.1
OS: Red Hat Linux 7.2

When "covmat" is supplied in princomp(), the output value
"center" is all
NA's, even though the input matrix was indeed centered. I haven't read 
anything about this in the help file for princomp(). See code below for an 
example: pc2$center is all NA's.

Jerome Asselin

x <- rnorm(6)
y <- rnorm(6)
X <- cbind(x,y)
pc1 <- princomp(X)
pc1$scores %*% t(pc1$loadings)-X
pc1$center
pc2 <- princomp(X,covmat=cov(X)*5/6)
pc2$scores %*% t(pc2$loadings)-X
pc2$center  # All NA's (should be the same as pc1$center)
> x <- rnorm(6)
> y <- rnorm(6)
> X <- cbind(x,y)
> pc1 <- princomp(X)
> pc1$scores %*% t(pc1$loadings)-X
               x          y
  [1,] 0.1522791 -0.4205906
  [2,] 0.1522791 -0.4205906
  [3,] 0.1522791 -0.4205906
  [4,] 0.1522791 -0.4205906
  [5,] 0.1522791 -0.4205906
  [6,] 0.1522791 -0.4205906
> pc1$center
         x          y
-0.1522791  0.4205906
> pc2 <- princomp(X,covmat=cov(X)*5/6)
> pc2$scores %*% t(pc2$loadings)-X
               x          y
  [1,] 0.1522791 -0.4205906
  [2,] 0.1522791 -0.4205906
  [3,] 0.1522791 -0.4205906
  [4,] 0.1522791 -0.4205906
  [5,] 0.1522791 -0.4205906
  [6,] 0.1522791 -0.4205906
> pc2$center  # All NA's (should be the same as pc1$center)
[1] NA NA

-- 

Jerome Asselin (Jérôme), Statistical Analyst
British Columbia Centre for Excellence in HIV/AIDS
St. Paul's Hospital, 608 - 1081 Burrard Street
Vancouver, British Columbia, CANADA V6Z 1Y6
Email: jerome@hivnet.ubc.ca
Phone: 604 806-9112   Fax: 604 806-9044

On Fri, 8 Aug 2003 jerome@hivnet.ubc.ca wrote:
> R version: 1.7.1
> OS: Red Hat Linux 7.2
> 
> When "covmat" is supplied in princomp(), the output value
"center" is all
> NA's, even though the input matrix was indeed centered. I haven't
read
> anything about this in the help file for princomp(). 
You *did* try to read the help page?  You seem to have failed to read the 
description of `value', which says

  center: the means that were subtracted.
> See code below for an  example: pc2$center is all NA's.
> x <- rnorm(6)
> y <- rnorm(6)
> X <- cbind(x,y)
> pc1 <- princomp(X)
> pc1$scores %*% t(pc1$loadings)-X
> pc1$center
> pc2 <- princomp(X,covmat=cov(X)*5/6)
If you had read either the help page or the code, you would have seen
that x is not used when you supply covmat.  So although means were
subtracted, they were done by you not princomp(), and NA is the right
answer.

-- 
Brian D. Ripley,                  ripley@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