Not enough input arguments error in nlinfit

Hello all,

I'm quite new to matlab and I'm trying to fit a non linear model to my data (I have 3 independent variables, 4000+ data points)

I have written the following code after following few you tube videos and matlab documents. Result is a 4000*4 matrix that has my three independent variables and the dependent variable

x=Result.x;% independent variable I

y=Result.y;% independent variable II

z=Result.z;% independent variable III

p=Result.fx; % dependent variable

inputs=[x,y,z];

my_fun = @(beta,x,y,z) (x.^beta(1) * y.^beta(2) / z.^beta(3)); % assumed model

initials = [1,2,1];

new_coeff = nlinfit(inputs,p,my_fun,initials);

However, I get the following error when I run this code

Error using nlinfit (line 205)
Error evaluating model function '@(beta,x,y,z)(x.^beta(1)*y.^beta(2)/z.^beta(3))'.

Error in Untitled5 (line 11)
new_coeff = nlinfit(inputs,p,my_fun,initials);

Caused by:
    Not enough input arguments.

Can someone help me? Why does it say not enough input parameters?

Answer 1

[Star Strider]

Try this:

inputs = [x(:),y(:),z(:)];
my_fun = @(beta,inputs) (inputs(:,1).^beta(1) .* inputs(:,2).^beta(2) ./ inputs(:,3).^beta(3)); % assumed model

That should work with this nlinfit call:

new_coeff = nlinfit(inputs,p,my_fun,initials

(I cannot test your code.)

[Awanthika Senarath]

Thanks, this was the error. I had passed too many arguments. Matlab seriously needs to re-do their error messages!

[Star Strider]

As always, my pleasure!

Answer 2

[Darren Wethington]

When passing arguments to nlinfit, you've (correctly) passed your inputs as one variable, "inputs". Now my_fun will take "initials" as the argument "beta", "inputs" as the argument "x", and... no input argument for "y" or "z".

Try changing my_fun to take two arguments, "beta" and "data" for instance. Then specify in my_fun that x=data(1,:), y=data(2,:), etc (or whatever makes sense for your data). Hopefully this helps.

Answer 3

[Awanthika Senarath]

I solved it, the model_fun only takes two parameters. I changed it as follows and now it works.

my_fun = @(beta,Results) (Result.x.^beta(1) .* Result.y.^beta(2) ./ Result.z.^beta(3));