0
$\begingroup$

Another beginner question: I'm trying to do PCA on compositional data. In other words, all the variables in the group add up to 100%.

I've since learned on this forum that compositional data poses a problem for linear regression. Will compositional data also pose a unique problem when doing PCA?

$\endgroup$
2
$\begingroup$

I'm in the same situation as you. PCA seems to have some difficulties for compositional data. I've listed some papers below that have been helpful to me. I'm still trying to understand everything, but these papers have a good overview I think. Or googling "Compositional Data" or "Compositional Data Principal Component Analysis" returns a lot of other papers that are freely available and informative.

Principal component analysis of compositional data J. Aitchinson

A Concise Guide to Compositional Data Analysis J. Aitchinson

Principal components analysis for compositional data with outliers P. Filzmoser, et al.

| improve this answer | |
$\endgroup$
0
$\begingroup$

The R package robcompositions has a function for PCA on compositional data. Details can be found here

http://rpackages.ianhowson.com/cran/robCompositions/man/pcaCoDa.html

| improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.