4

I would like to know the importance of the original features in principle component analysis. See this Stackoverflow link for an example of what I mean (with a code example).

The question is: can you multiply the factor loadings [meaning the corresponding eigenvector] of a PC by the explained variance of the PC to assess the importance of features in a PCA?

Community
  • 1
Guido
  • 141
  • 5
  • What is the purpose of the feature selection you are trying to do? – amoeba Mar 10 '17 at 14:47
  • Note that what you call "factor loadings" are simply PCA eigenvectors (I edited to clarify). "Loadings" is something else: http://stats.stackexchange.com/questions/143905 - namely eigenvectors scaled by the square roots of the respective eigenvalues. So you have basically reinvented the loadings. – amoeba Mar 10 '17 at 14:49
  • I would like to know which original features contributed most to the PCA distribution. In that way, I could better explain the PCA distribution – Guido Mar 15 '17 at 14:54

0 Answers0