Curve Fitting Toolbox with multiple variables

Question

Ahmet Hakan UYANIK 2022-6-7

回答： Torsten 2022-6-7

Hello,

I have generated the code with cftool of matlab however it seems like more than 2 independent variable is not possible with this toolbox.

There are some solutions provided like lsqcurvefit or lsqnonlin but they are not limiting range of coefficients as generated code from cftool below.

Do you know how can i apply with 3 independent variables with limiting the range of coefficients as below?

Thank you in advance for your comments and ideas

[xData, yData, zData, sData] = prepareSurfaceData( Wm, Ym, Qm, Sm);

% Set up fittype and options.

ft = fittype( ['a*x^b*s^c*y^d'], 'independent', {'x', 'y','s'}, 'dependent', 'z' );

opts = fitoptions( 'Method', 'NonlinearLeastSquares' );

opts.Display = 'Off';

opts.Lower = [-Inf 0.9 0.3 1.3];

opts.Upper = [Inf 1.1 0.7 1.8];

opts.StartPoint = [0.706046088019609 0.0318328463774207 0.27692298496089 0.0461713906311539];

Answer 1

Torsten 2022-6-7

lsqnonlin and lsqcurvefit allow to prescribe lower and upper bounds (lb,ub) on the parameters: