Re: confidence region

[Henry F. Mollet]

Check section 'set return value' in 'leasqr.m to find the definition of Z: Y
Z = (m-n) * jac' * Qinv * jac / n * resid' * Qinv * resid).
(jac is the Jacobian)

In linear case this translates to according to my annotations without
guarantee and using p for n, n for m, X for jac, and X'*X for jtgjinv)
Z = (n-p) /(p resid' * resid) * X'*X = 1/(p * MSE) * (X'*X).

My annotations also give
Z = 1/ (p MSE^2) * acovp (alternate covariance matrix of parameters).
Therefore Z-matrix is related to the reciprocal of the alternate covariance
matrix.

It's best to try an example and I'll check if I can still do it.
Henry


on 2/17/06 4:43 AM, Janssens Job at address@hidden wrote:

> Hello 
>  
> I am a student using leasqr.m for data fitting, but I am having trouble
> interpreting the matrix Z, which is described as "the matrix that
> defines the confidence interval". What does this exactly mean, and what
> do I have to do to know the uncertainties on my parameters?
>  
> Kind regards,
> Job Janssens.
> 
> 
>   ----------  SCK-CEN Disclaimer  ---------
>   http://www.sckcen.be/emaildisclaimer.html
> 




-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------