defining a fitting type

I have written a funcion called Even_fit.m and there are 5 coefficients that need to be found from the fit (a,b,c,d,f) and there is one variable N that I want to load from the workspace, which is already fixed before fitting. My question is how I can load this variable in the fit, since when I do:

ft_even = fittype("Even_fit(x,a,b,c,d,f,N)");

It will also take N as a fitting coefficient, which I do not want.

Thank you!

Answer 1

[Torsten]

My guess is

fitType = @(x,a,b,c,d,f)Even_fit(x,a,b,c,d,f,N);
fit(x,y,fitType)

[Wout Laeremans]

This does not work

[Torsten]

Put the vector of the independent variable to the end of the input list for your function "EvenFit":

x = linspace(0,1,10).';
y = 3*x.^2 + 0.1*rand(10,1);
n = 12;
fitType = @(a,b,c,x)Even_fit(x,a,b,c,n);
sol = fit(x,y,fitType)

Warning: Start point not provided, choosing random start point.

sol = 
     General model:
     sol(x) = Even_fit(x,a,b,c,n)
     Coefficients (with 95% confidence bounds):
       a =       2.803  (2.517, 3.09)
       b =      0.2072  (-0.09066, 0.505)
       c =    0.009327  (-0.05463, 0.07328)

plot(sol,x,y)

function yfit = Even_fit(x,a,b,c,n)
  yfit = a*x.^2+b*x+c;
end

Answer 2

[Steven Lord]

See the "Create Custom Nonlinear Models and Specify Problem Parameters and Independent Variables" and "Use Anonymous Functions to Work with Problem Parameters and Workspace Variables" examples on the fittype documentation page.