Hi all,
I have a data set that I want to fit a curve. In order to determine which function agrees with my data. I need the coefficient of determination that can indicate how well it fits. But that coefficient of determination should be able to be used for nonlinear and linear fit because I am testing multiple models and want to compare how well each one fits.
Could you share a coefficient of determination that could be used for both nonlinear and linear regression?
Also, on a side note, why does coefficient of determination not work for nonlinear regression?
Thanks!
Hi there,
if you check the wiki (https://en.wikipedia.org/wiki/Coefficient_of_determination) you can see that for the nonlinear regression, the partition of the sum of squares doesn't necessary hold.
However, you can still compute R2 as follows for you nonlinear models and compare it with the linear ones.
R2 = 1 - (SSres)/(SStot)
- Badges
- Science topic
- Similar topics
- Mathematics