1

I bet I had seen a paper which used PCA and obtained a "curious" result, only later it was discovered that they didn't standardize the variables before the PCA.

It had to do with the weather if I am not mistaken.

Do you know that paper, or know some examples of papers in which PCA was misused in this way, without standardizing the variables first?

robertspierre
  • 1,358
  • 6
  • 21
  • 3
    It's not necessarily a misuse. If the variables have the same measurement units and in the given application a larger variance means that the variable is more informative, one should *not* standardise, as this would lose the information in the variance. It's by the way very unlikely that somebody knows what paper you mean based on the thin information given. – Christian Hennig Jun 26 '21 at 19:56

0 Answers0