[R] About princomp

[Prof Brian Ripley]

You send HTML mail (please see the posting guide and do as we ask) and it 
seems to have not produced a working text version.

My guess is that you have a changed version of USArrests around.  In any 
case, there is no problem in R 2.1.1 (sic), so this is local to you.

On Fri, 16 Sep 2005, Matthieu Cornec wrote:

> Hi,
> I run the example for princomp for R211
> I got the following error for biplot
>
>> ## The variances of the variables in the
>> ## USArrests data vary by orders of magnitude, so scaling is appropriate
>> (pc.cr <http://pc.cr> <- princomp(USArrests)) # inappropriate
> Erreur dans cov.wt(z) : 'x' must contain finite values only
>> princomp(USArrests, cor = TRUE) # =^= prcomp(USArrests, scale=TRUE)
> Erreur dans cov.wt(z) : 'x' must contain finite values only
>> ## Similar, but different:
>> ## The standard deviations differ by a factor of sqrt(49/50)
>>
>> summary(pc.cr <http://pc.cr> <- princomp(USArrests, cor = TRUE))
> Erreur dans cov.wt(z) : 'x' must contain finite values only
>> loadings(pc.cr <http://pc.cr>) ## note that blank entries are small but
> not zero
>
> Loadings:
> Comp.1 Comp.2 Comp.3
> Murder 0.663 0.314 0.679
> Assault 0.694 -0.715
> UrbanPop 0.279 -0.946 0.165
>
> Comp.1 Comp.2 Comp.3
> SS loadings 1.000 1.000 1.000
> Proportion Var 0.333 0.333 0.333
> Cumulative Var 0.333 0.667 1.000
>> plot(pc.cr <http://pc.cr>) # shows a screeplot.
>> biplot(pc.cr <http://pc.cr>)
> Erreur dans plot.window(xlim, ylim, log, asp, ...) :
> 'xlim' nécessite des valeurs finies
>>
>
> 	[[alternative HTML version deleted]]
>
>

-- 
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