Two recent questions:

I am sure we have had this question before, but a simple search did't show anything that seemed useful, nor was there anything among the "related" threads.

Do we have a canonical duplicate for this kind of questions? If not, should we have one?

| |
  • 2
    $\begingroup$ "Why does my predictive model explain less than 100% of the variance?" is IMHO too general for a canonical question. $\endgroup$ – amoeba Feb 7 '19 at 16:11
  • 4
    $\begingroup$ @amoeba: and it's a different question. $\endgroup$ – Stephan Kolassa Feb 7 '19 at 16:13
  • 1
    $\begingroup$ Hmm, why is it different? $\endgroup$ – amoeba Feb 7 '19 at 17:28
  • 2
    $\begingroup$ Maybe a general question on "what is regression to the mean, and what does it have to do with linear regression"? $\endgroup$ – Matthew Drury Feb 8 '19 at 1:34
  • 6
    $\begingroup$ @amoeba: the difference is that the questions in question deal with out-of-sample predictions, whereas explained variance deals with in-sample fits. (Yes, I'll admit the two concepts are related. But I don't think someone posting the first kind of question will understand how an answer to the second kind will apply to their question.) $\endgroup$ – Stephan Kolassa Feb 8 '19 at 6:56

You must log in to answer this question.

Browse other questions tagged .