I have never seen this term mentioned before. Yet this study uses it: https://www.econstor.eu/bitstream/10419/204328/1/ifro-wp-2011-12.pdf

Is it any different than the typical R^2, i.e. coefficients of determination?


Yes, they are variants of the classical $R^2$.

  • McElroy coefficient of determination ${R^2_M}$ measures the goodness–of–fit in a system-of-equations model(several regression equations instead of one as in the classical case)
  • coefficient of determination of estimation ${R^2_e}$ is the classical $R^2$.
  • coefficient of determination of prediction${R^2_p}$ is classical $R^2$ evaluated on the single observation that was omitted during LOOCV procedure.

I hope it helps!!

| cite | improve this answer | |
  • $\begingroup$ thank you. Could you provide any source or additional info on the coefficient of determination of prediction? (Btw, did you mean "evaluating upon removing a single observation during a LOCV procedure) $\endgroup$ – StatsScared Feb 18 at 4:18
  • 1
    $\begingroup$ The paragraph "This procedure repeats the statistical analysis N times, where N is the number of observations and at" in the paper you cited states that it uses LOOCV to facilitate the evaluation of the system of regression equations. As for the reference, sciencedirect.com/science/article/abs/pii/0304407677900082. Hope this helps:) $\endgroup$ – Nizam Feb 18 at 4:59

Your Answer

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

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