least squre data fit for a function of two variables

z=f(x,y) is my function.but i know data points (x ,y) and corresponding z values only. I know that z is of the form z=a*g(x,y)+b*h(x,y) i want find a,b in a least square sense. Can i use 'lsqcurvefit' function?? If i can't what can i use?? note: Z is not linear

for example

xdata = [1 4;2 5;3 6];%%f(1,4)=11    f(2,5)=2      f(3,6)=63
ydata = [11;32;63];
 x = lsqcurvefit(@myfunction, [1 7], xdata, ydata);

function f=myfunction(x,xdata)
f = f = x(1)*(xdata(:,1)*xdata(:,2))+x(2)*(xdata(:,1))^2;

is not working

Answer 1

[David Sanchez]

Use the curve fitting toolbox

cftool

take a look at

doc cftool

Answer 2

[Matt J]

note: Z is not linear

Z is linear as a function of a and b and that's the only dependence that matters. You don't need any fancy solvers:

 A=[xdata(:,1).*xdata(:,2),  (xdata(:,1)).^2];

 x=A\ydata,