Programmatically calculate goodness of fit using Curve Fitting Toolbox?

Question

Tim 2014-2-8

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/115337-programmatically-calculate-goodness-of-fit-using-curve-fitting-toolbox

评论： Tim 2014-2-10

Hi,

I am hoping to retrieve the goodness of fit values that the curve fit tool calculates for a custom model, but programatically. My model is:

g=fittype('a*sin(2*pi*25000000*x+b)+c')

which I fit to x and y data with:

f=fit(x,v,g)

(where v is for voltage). This returns the following:

f = 
       General model:
       f(x) = a*sin(2*pi*25000000*x+b)+c
       Coefficients (with 95% confidence bounds):
         a =      -13.03  (-13.04, -13.02)
         b =      -4.746  (-4.747, -4.745)
         c =     -0.3985  (-0.4073, -0.3897)

From here I want to see the sum of squares due to error (SSE), R-square, Adjusted R-square, and Root Mean Square Error (RMSE), as a quantitative assessment of the model quality. Does anyone have any solutions? There is no obvious way of retrieving this information listed in the documentation for the toolbox.

Thanks in advance.