[ADMB Users] Objective function value in robust regression
Rubén Roa
rroa at azti.es
Mon May 17 00:34:20 PDT 2010
> -----Mensaje original-----
> De: users-bounces at admb-project.org
> [mailto:users-bounces at admb-project.org] En nombre de dave fournier
> Enviado el: viernes, 14 de mayo de 2010 9:22
> Para: users at admb-project.org
> Asunto: Re: [ADMB Users] Objective function value in robust regression
>
> I guess the obvious thing is to minimize a function where you
> know the answer and see what you getin the par file. Easier
> than looking at the source.
> Say
>
> f=square(x(1)-1.0)+square(x(1)+1.0);
>
> If answer if f=2 you have -LL and if you get f=-2 you have the LL.
Thanks.
I just thought that since you provided this fitting technique (robust regression) pre-packaged for the user (the user did not have to write down the negative log-likelihood), then you might have been so kind to also produce the log-likelihood value in the par file, instead of just printing the value of the objetive function value (the negative log-likelihood). Si I was wondering if I should reverse the sign of the printed objective function value to compute the AIC for the particular case of the robust regression.
____________________________________________________________________________________
Dr. Rubén Roa-Ureta
AZTI - Tecnalia / Marine Research Unit
Txatxarramendi Ugartea z/g
48395 Sukarrieta (Bizkaia)
SPAIN
More information about the Users
mailing list