how to use nlmefit to multilevel regression (mixed effect) model

I have the following data:

X - column vector of n predictor values

Y - column vector of n responses

G - column vector of n values defining group membership in one of g groups

The model to be fitted is

y_ij = b_0 + b_1*.x_i + u_j

The values y_ij are presented in the vector Y for the corresponding values of the vector X.

How can I arrange the calling the nlmefit function, especially FUN parameter.

Answer 1

[the cyclist]

This answer

http://www.mathworks.com/matlabcentral/answers/44620-group-specific-predictors-in-nlmefit

has a toy example that I think will be helpful for you.

[Dimitar Atanasov]

If fact it is not my case, or I cant figure out how to rearrange the things.

In my model

y_ij = b_0 + b_1*.x_i + u_j

b_0, b_1 and u_j (j = 1:g) are unknown parameters.

[the cyclist]

I'm afraid I can't code up your specific problem for you. I was hoping that the simple example that Tom posted would give you a hint.

There is also a thorough example in the documentation of nlmefit(). Have you looked it and tried to understand it?

http://www.mathworks.com/help/stats/nlmefit.html

[Dimitar Atanasov]

Yes, of course. The first place that e look for solution was the documentation....

I'll try to manage the problem and I'll post the solution after that.