[R] help with PCA and classical scaling

[Christian Hennig]

Hi Domenico,

multidimensional scaling methods (functions cmdscale, isoMDS, sammon)
start with a dissimilarity matrix while PCA starts from a data
matrix with variables. Equivalence of PCA and classical MDS refers to
the situation where you use the Euclidean distances between the data
points on which you apply PCA. (Note: I cite this from my memory, so check
it before believing it.)

> My first question is: Do I need an initial embedding of my data before
> applying PCA through the methods princomp or prcomp in the stats package?

I don't think that you need PCA at all (depends on what exactly you want
to do) - an MDS method alone could be enough for you
(cmdscale, if you want to do something analogous to PCA).

Best,
Christian



*** NEW ADDRESS! ***
Christian Hennig
University College London, Department of Statistical Science
Gower St., London WC1E 6BT, phone +44 207 679 1698
chrish at stats.ucl.ac.uk, www.homepages.ucl.ac.uk/~ucakche