0
$\begingroup$

If I have a PCA that I ran on some set of variables, how (if at all) will it relate to the PCA results if I add or remove one variable?

Will the PCA components change in some well-defined way, or is there no relationship that is guaranteed between the old and new eigenvalues & eigenvectors?

$\endgroup$
  • $\begingroup$ The variable you mean here is the number of observation or the dimension of the data? $\endgroup$ – Vickyyy Mar 17 '18 at 6:32
  • $\begingroup$ The dimension of the data $\endgroup$ – Thomas Johnson Mar 18 '18 at 3:38
0
$\begingroup$

There is no guarantee - it depends on the distribution of the added / removed variable.

$\endgroup$
  • $\begingroup$ This does not provide an answer to the question. To critique or request clarification from an author, leave a comment below their post. - From Review $\endgroup$ – kjetil b halvorsen Mar 18 '18 at 13:05
  • $\begingroup$ @kjetilbhalvorsen The question was whether there is a guaranteed relationship between the old and new PCA. The answer is that there is no guaranteed relationship. This is not a critique, it is simply the answer. $\endgroup$ – elliotp Mar 18 '18 at 14:07

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.