How do I use the curve fitting tool with a custom function that does not depend of x?

I always get an error in the toolbox when I try to fit to a Matlab function I have. I have a set of points to which I want to fit some data. Problem is, the points are, for example, for x=1.4, 2.5, 4.1,... , and the data I have is for x=1, 2, 3,... , yet the way this data is calculated does not dependent of the value of x, even though each individual data point is assigned to x=1, 2, 3,.... I think that the fact that it does not depend of x must be the problem, since when I use this function and just assign some values to the parameters f1,f2,f3,f4,f5,a,b,c,d I obtain a fit (though wrong unless I am able to manually fit the parameters).

If I just use fit(f1,f2,f3,f4,f5,a,b,c,d) as the custom equation I get the error:

"The independent variable x does not appear in the equation expression."

And I use fit(f1,f2,f3,f4,f5,a,b,c,d,x) I get:

"Expression fit(f1,f2,f3,f4,f5,a,b,c,d,x) is not a valid MATLAB expression, has non-scalar coefficients, or cannot be evaluated"

What could be the problem? The function I have is as follows:

function f=fit(f1,f2,f3,f4,f5,a,b,c,d)

x=(1:(size(stopProb,2)-1))';

sum1=1:(size(stopProb,2)-1); 
sum2=1:(size(stopProb,2)-1);
sum3=1:(size(stopProb,2)-1);
sum4=1:(size(stopProb,2)-1);
sum5=1:(size(stopProb,2)-1);

for i=2:size(stopProb,2)
    sum1(i-1)=sum(stopProb(1:a,i));
    sum2(i-1)=sum(stopProb(a+1:b,i));
    sum3(i-1)=sum(stopProb(b+1:c,i));
    sum4(i-1)=sum(stopProb(c+1:d,i));
    sum5(i-1)=sum(stopProb(d+1:size(stopProb,1),i));
end
v=[sum1;sum2;sum3;sum4;sum5];
sumTotal=v';

probInterval=zeros(size(sumTotal,1),2);
f_dia_fit=zeros(size(sumTotal,1),1);

for u=1:size(sumTotal,1)
    probInterval(u,1)=sumTotal(u,1)/(sumTotal(u,1)+sumTotal(u,2)+sumTotal(u,3)+sumTotal(u,4)+sumTotal(u,5));
    probInterval(u,2)=sumTotal(u,2)/(sumTotal(u,1)+sumTotal(u,2)+sumTotal(u,3)+sumTotal(u,4)+sumTotal(u,5));
    probInterval(u,3)=sumTotal(u,3)/(sumTotal(u,1)+sumTotal(u,2)+sumTotal(u,3)+sumTotal(u,4)+sumTotal(u,5));
    probInterval(u,4)=sumTotal(u,4)/(sumTotal(u,1)+sumTotal(u,2)+sumTotal(u,3)+sumTotal(u,4)+sumTotal(u,5));
    probInterval(u,5)=sumTotal(u,5)/(sumTotal(u,1)+sumTotal(u,2)+sumTotal(u,3)+sumTotal(u,4)+sumTotal(u,5));
    f_dia_fit(u,1)=probInterval(u,1)*f1+probInterval(u,2)*f2+probInterval(u,3)*f3+probInterval(u,4)*f4+probInterval(u,5)*f5;
end

f=f_dia_fit;

[Stephan]

What error do you get? please show us the complete red text.

without knowing more, it can be said that you should rename your function. fit is the name of an inbuilt function, which will cause problems.

[Walter Roberson]

In this context, x is the independent variable, which does not need to literally be named x.

If your values do not depend upon your independent variables, then you have two possibilities:

1. Your values are constant. There is no useful fitting that can be done in this case.
2. Your values are random. There is no meaningful fitting that can be done in this case.

[André Simões]

You just made me realize that I did not express myself well. What I meant is that to calculate the values of y of this data are not calculated using a x value, like for example a linear regression (y=mx+b), yet they do correspond to some x values (x=1, 2, 3,...). which is why I write

x=(1:(size(stopProb,2)-1))

in the function.

[Walter Roberson]

Think of using t as the independent variable if that helps.

The first time step, t=1, you get a value which happens to be stored in y(1)

The second time step, t=2, (a fixed interval later), you get a value which happens to be stored in y(2)

The third time step, t=3, (the fixed interval after the second), you get a value which happens to be stored in y(3)

and so on.

It does not matter that you have no literal 'x' or 't' in your sequence of values: it only matters that they represent function values at consistent step size. In such a case, you can use the index as your independent variable. x = 1 : length(y) is fine.

Answer 1

[André Simões]

There is no red text since I am using the Curve Fitting Tool. I have changed the function name. The errors I get are those I described.

If I just use metodo_fit_v3(f1,f2,f3,f4,f5,a,b,c,d) as the custom equation I get the error:

"The independent variable x does not appear in the equation expression.

Use x in the expression or indicate another variable as the independent variable."

If I use metodo_fit_v3(f1,f2,f3,f4,f5,a,b,c,d,x) I get:

"Expression metodo_fit_v3(f1,f2,f3,f4,f5,a,b,c,d,x) is not a valid MATLAB expression, has non-scalar coefficients, or cannot be evaluated:

Error in fittype expression ==> metodo_fit_v3(f1,f2,f3,f4,f5,a,b,c,d,x)

??? Undefined function 'metodo_fit_v3' for input arguments of type 'double'."