Effacer les filtres
Effacer les filtres

Curve fitting toolbox doesn't give confidence intervals

7 vues (au cours des 30 derniers jours)
Edoardo_a
Edoardo_a le 22 Mar 2023
Réponse apportée : Drew le 16 Nov 2023
Hi all, I have a problem with the curve fitting toolbox program.
I did make a code out of the toolbox curve fiitting because I needed to modify the parameters in a specific way for different curves that I am fitting with a sum of Gaussians and Lorentzian. It seems to work well but the problem is that for some reason, some times the fittin doesn't show the confidence intervals for specific parameters and I can't understand why: Does anyone know how to help me?
Thanks in advance, I update some files to make it more clear.
  2 commentaires
Matt J
Matt J le 22 Mar 2023
We would need to see a demonstration of what you mean.
Edoardo_a
Edoardo_a le 23 Mar 2023
Hi Matt, maybe from a more generic point of view: why when changing the initial parameters in the curve fitter app (using a custom equation), for same values the fitting doesn't give any confidence interval for some parameters?(as I shhown in the ex1.txt file I uploaded)
Even if the fit is good I dont't understand why doesn't show any confidence interval

Connectez-vous pour commenter.

Réponses (1)

Drew
Drew le 16 Nov 2023
See https://www.mathworks.com/help/curvefit/cfit.confint.html, where it says: "If coefficients are bounded and one or more of the estimates are at their bounds, those estimates are regarded as fixed and do not have confidence bounds."
Notice the message "(fixed at bound)" in the "ex1.txt" file that you attached. So, it looks like the coefficients that are fixed at their bound is the reason for not returning confidence intervals. Therefore, if the coefficient bounds are relaxed, the algorithm may be able to successfully return confidence intervals.
If this answer helps you, please remember to accept the answer.

Community Treasure Hunt

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

Start Hunting!

Translated by