0
$\begingroup$

Which components should use for plotting in a PCA analysis? Should it be component 1 versus component 2, or any combination that shows clustering is okay to use?

Also, I have seen that in a few cases the axis labels mention the variance that is shown (e.g. it says "Principal component 1 (Var. 58.09%)"). Does this mean that in these plots only a part of the corresponding axis is shown (i.e. is it zoomed-in)?

Improve this question
$\endgroup$
2
  • 2
    $\begingroup$ Your questions bely a fundamental lack of knowledge of what PCA does. I strongly recommend you read the following article. If you feel comfortable with linear algebra and basic statistics, you can skip to page 13. The meat of the discussion is only 7 pages with graphics, and is the most accessible explanation of PCA I've encountered: cs.otago.ac.nz/cosc453/student_tutorials/… . I also suggest you avoid the temptation to use tools you don't understand in general, since you probably won't interpret their output correctly. $\endgroup$
    – David Marx
    Nov 15, 2013 at 16:17
  • $\begingroup$ Thanks for the link, David! I'll definitely have a look. $\endgroup$
    – panos
    Nov 18, 2013 at 10:48

1 Answer 1

Reset to default
1
$\begingroup$

You do not need to show any at all. You should show those that are instrumental for the argument you are making/pieces of evidence you are interested in.

The "var" indicates that the component plotted accounts for so-and-so many % of the variance among variables included.

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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