Fitting with nlinfit

Hi

I am fitting a function to some data using nlinfit in the following way

[b, r, J, COVB, mse] = nlinfit(v(:, 1), v(:, 2), model, b1, statset('Display','iter'));

My problem is that I want to find the uncertainties on the fitted parameters given in b1. Is it correct that the uncertainties on the final parameters in "b" are simply the square root of the diagonal elements of COVB?

Thanks in advance.

Best, Niles.

Answer 1

[the cyclist]

Yes. The diagonal elements of COVB are the (estimated) variances of each coefficient distribution, and their square roots are the standard deviations. So, assuming by "uncertainty" you mean one standard deviation, then you are correct.