MATLAB: Workaround on curvefit limitation

curve fittingfourier

Hello,

Is there any workaround to fit a function using the fit command without the limitation of 8 coefficients (considering the sum of sines or Fourier, specifically)?

Examples: model = fit(x,y,’sin10’) or model2 = (x,y,’fourier11’).

Thanks in advance.

Best Answer

The built-in library of fits only contains sum of sines and Fourier fits up to sin8 and fourier8. If appropriate you could write your own custom fit using the fittype function, but keep in mind the advice from Jeff Goldblum's character in Jurassic Park: "Your scientists were so preoccupied with whether they could that they didn’t stop to think if they should."

Don't fit a sum of 30 sine functions (for example) just because you can, ask yourself if fitting a sum of 30 sine functions is a reasonable thing to do based on the problem you're trying to solve.

Related Solutions

MATLAB: Percent residual

You could save your fitting values to workspace, using cftool choose analysis and there select the option evaluate fit at Xi (in your case Xi is your x) . Now if you save those results you 'll have a structure containing xi and yfit. You can calculate percent residuals r(%)=100*(yi-yifit)./yi. Just be careful in case your original data aren't sorted by x value to sort them because the output results of fitting analysis are sorted to xi.

MATLAB: RMSE

One way is to compute it yourself. You just need to compute the root of the mean of the squared errors (hence the name):

y = <true values>
yhat = <fitted values>
RMSE = sqrt(mean((y - yhat).^2));

Best Answer

Related Solutions

MATLAB: Percent residual

MATLAB: RMSE

Related Question