Date: Fri, 10 May 2002 09:22:36 -0700
Reply-To: Dale McLerran <stringplayer_2@YAHOO.COM>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Dale McLerran <stringplayer_2@YAHOO.COM>
Subject: Re: help for maximum likelihood
Content-Type: text/plain; charset=us-ascii
The procedure NLMIXED is excellent for maximum likelihood estimation
of arbitrary likelihood functions. Just yesterday I posted to the
list a maximum likelihood solution for a zero-inflated Poisson
distribution which has a simple solution in NLMIXED. If you want
to look for that posting, it has a subjet heading of "Lognormal".
I have also posted within the past couple of months a solution for
a positive Poisson model. Interesting that these models are both
Poisson, but one has an excess of zero values while the other has
a deficit of zero values. If you go to the link below and search
for NLMIXED, you will find a number of postings on the subject.
One distribution which the NLMIXED procedure does seem to have
difficulty fitting (or at least that I have had trouble getting to
converge properly) is the negative binomial distribution. But the
GENMOD procedure does properly implement a negative binomial
solution, so it does not concern me too greatly that I could not
get the NB to converge in NLMIXED.
Hope this gets you started.
--- hui li <lihui@UNM.EDU> wrote:
> Hello all,
> As the beginner to SAS, I always use canned procedure for MLE, like
> logistic, probit... I am wondering what is the good way to maximize
> forms of likelihood function? i tried Proc NLP, it is so hard to get
> converge, i also tried Proc IML. Could anybody give me some advice on
> "long-hand"maximum likelihood estimation programming?
> Hui Li /~~~~\ /~~~~\
> Grad. student,UNM \ \ / /
> Dept. of Economics \ \ / /
> Phone(Office):505-277-1953 \ / \ /
> Email: email@example.com / ----- \
> / @ @ \
> \ 0 /
> ~~~~~~~ ~~~~~~~
Fred Hutchinson Cancer Research Center
Ph: (206) 667-2926
Fax: (206) 667-5977
Do You Yahoo!?
Yahoo! Shopping - Mother's Day is May 12th!