Is the 8th term maximum in the curve fitting toolbox and command line?

It seems to apply up to 9th term with a simple command line input.

From 'sin10', it is recognized as 3D. is there any other way?

I hope for your answer. thank you.

F = fix(x,y,'sin9')
plot(F,x,y)

Answer 1

[Matt J]

You will probably have to use a custom model, or else resort to lsqcurvefit().

[재열 김]

Thank you for answer!

I used a custom model, but got worse results than using the sum of sine model as a startpoint.

Is there a way to get an optimal startpoint?

[Jongil Lee]

Hi,

There're some options that you mentioned as "optimal startpoint" as below,

모수적 피팅 - MATLAB & Simulink - MathWorks 한국

"Sum of Sines Models" supports up to 8th order.

[재열 김]

@Jongil Lee

Thank you for answer. I'd like to get a startpoint that is optimized equal to the sum of sine model.

However, the startpoint for a custom nonlinear model is arbitrary.

Of course I can input the coefficients directly, but I can't find the appropriate coefficient values.

So I want to know how to get an optimized startpoint.

thanks!

[Matt J]

For a startpoint, perhaps do a non-custom 'sin8' fit to estimate the first 24 parameters. Subtract the sin8 fit from the measured curve samples to obtain a a modified set of curve samples to which you would do a non-custom 'sin2' fit to estimate the remaining parameters.

F=fit(x,y,'sin8');

dy=y-F(x);

dF=fit(x,dy,'sin2');

startpoint=[coeffvalues(F),coeffvalues(dF)]; %startpoint for custom sin10 fit

Answer 2

[Matt J]

Perhaps just do an FFT decomposition and select the 10 strongest frequencies.