How can I find the best fit to a bunch of data?

Hi,

I am trying to find the best fitting to a bunch of data. I keep getting an error as below:

*x = lsqcurvefit(@fcn,[.2, .2, .2] ,[T, V, M] , ones(93, 1)); ??? Error using ==> lsqcurvefit at 253 Function value and YDATA sizes are incommensurate.*

Below are my m-files:

x = lsqcurvefit(@fcn,[.2, .2, .2] ,[T, V, M] , ones(93, 1));

function out = fcn(param, x) alpha = param(1); beta = param(2); gamma = param(3); T = x(1); V = x(2); M = x(3); out =(T/550)^alpha + (V/100)^beta + (M/30)^gamma;

Any help will be appreciated.

Answer 1

I'm assuming the lengths of T,V, and M are each 31. If so, you probably meant to do this:

 x = lsqcurvefit(@fcn,[.2, .2, .2] ,[T,V,M] , ones(31, 1));

    function out = fcn(param, x) 
       alpha = param(1); 
       beta = param(2); 
       gamma = param(3); 

       T = x(:,1); V = x(:,2); M = x(:,3); 

     out =(T/550)^alpha + (V/100)^beta + (M/30)^gamma;

Hi Matt,

Thank you for your answer. T, V, and M are each 101. They are 101*1 vector each. I have corrected based on what you did and below is the new errors:

*??? Error using ==> mpower Inputs must be a scalar and a square matrix. To compute elementwise POWER, use POWER (.^) instead.

Error in ==> fcn at 9 out = (T/550)^alpha + (V/100)^beta + (M/30)^gamma;

Error in ==> lsqcurvefit at 209 initVals.F = feval(funfcn_x_xdata{3},xCurrent,XDATA,varargin{:});

Error in ==> TMV at 1 x = lsqcurvefit(@fcn,[.2, .2, .2] ,[T, V, M] , ones(101, 1)); Caused by: Failure in initial user-supplied objective function evaluation. LSQCURVEFIT cannot continue.*

Regards, Amir

well then, do as the error suggests, as follows

 out =(T/550).^alpha + (V/100).^beta + (M/30).^gamma;

Thank you, It seems that it works now with the below comment:

Local minimum possible. lsqcurvefit stopped because the size of the current step is less than the default value of the step size tolerance. criteria details * *Optimization stopped because the norm of the current step, 6.612031e-008, is less than options.TolX = 1.000000e-006.

Optimization Metric Options norm(step) = 6.61e-008 TolX = 1e-006 (default)* *

Any suggestion?

Amir,

See if the fit is good. If not, try to find a better initial guess.