creating biplots from existing multivariate output
I have a question about creating biplots from *existing* output by PCA, canonical variates analysis (CVA) and correspondence analysis. Am I right in thinking that variable and object scores in correspondence analysis are scaled to the same space and so require no further treatment for producing biplots? Where can I find a clear account that would help me calculate biplot values from existing output, bearing in mind (a) the various definitions of biplots, and (b) the various treatments that different packages give to PCA and CVA output (e.g. eigenvectors or variable loadings in PCA). = Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at http://jse.stat.ncsu.edu/ =
Re: creating biplots from existing multivariate output
>I have a question about creating biplots from *existing* output by >PCA, canonical variates analysis (CVA) and correspondence analysis. > Legendre & Legendre 1998, Numerical Ecology, 2nd ed, Elsevier, Amsterdam provides excellent descriptions of the different ways of displaying biplots and their interpretation. See also Gower, J. C. and D. J. Hand. 1996. Biplots. Chapman & Hall, London. 277 pp. Both books discuss Gabriel's original concept of the biplot and the several different ways biplots can be displayed. = Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at http://jse.stat.ncsu.edu/ =