getting very high MSE and an R almost equal to 1!!

4 vues (au cours des 30 derniers jours)
Med Aymane Ahajjam
Med Aymane Ahajjam le 2 Mar 2017
Commenté : Greg Heath le 21 Oct 2018
I am trying to model PV output power based on 6 parameters (components of solar irradiance, ambient temperature, wind speed, wind direction, timestamp). I have used the nftool but I have noticed that the values of MSE for training, validation and testing are very high, while R are always almost equal to one , as shown in the picture attached. Do you think I should change MSE parameters to normalized percent ? or something else to get better performance with accurate results..

Réponse acceptée

Greg Heath
Greg Heath le 7 Mar 2017
Your R looks pretty good. Therefore your MSE is relatively low compared to the average target variance because
R = sqrt( 1 - MSE/mean(var(t',1)) )
Thank you for accepting my answer
Greg
  2 commentaires
Med Aymane Ahajjam
Med Aymane Ahajjam le 7 Mar 2017
Thank you Greg for your answer.. I was wondering if I should compute another performance parameter more relative to this problem? any suggestions?
Greg Heath
Greg Heath le 21 Oct 2018
Yes.
Define a performance parameter that is more relative to the problem.
Then calculate it and post it.
Greg

Connectez-vous pour commenter.

Plus de réponses (0)

Catégories

En savoir plus sur Sequence and Numeric Feature Data Workflows dans Help Center et File Exchange

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by