Could anyone explain this to me? I couldn't find anything about it.

An example of the spiked covariance structure I have seen is a covariance matrix with zero non-diagonal elements and decreasing diagonal elements. Is this the general form of the spiked covariance structure?

  • $\begingroup$ Can you give a few detail? Like where you heard the term/ anything else on the internet that mentions it? $\endgroup$ Nov 4, 2021 at 3:09
  • $\begingroup$ @Tanner Phillips Thank you. I have added some details. $\endgroup$ Nov 4, 2021 at 5:09
  • $\begingroup$ My impression is that 'spiked' refers to the eigenvalues that are significantly above the value of random noise and appears to be related to scree plots and sparse PCA (where the term refers to setting low eigenvalues to 0, rather than PCA from sparse data). Without more concrete information it is hard to be sure I'm talking about the same concept. $\endgroup$
    – ReneBt
    Nov 4, 2021 at 5:49
  • $\begingroup$ @ReneBt There is no particular background in the paper where I came across the term. I will look into the sparse PCA:) $\endgroup$ Nov 4, 2021 at 6:07


Your Answer

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

Browse other questions tagged or ask your own question.