[R] glmmADMB: Mixed models for overdispersed and zero-inflated count data in R

[Hans Skaug]

Dear R-users,

Earlier this year I posted a message to this list regarding
negative binomial mixed models in R.  It was suggested that
the program I had written should be turned into an R-package.
This has now been done, in collaboration with David Fournier
and Anders Nielsen.

The R-package glmmADMB provides the following GLMM framework:
- Negative binomial or Poisson responses.
- Zero-inflation (optionally), e.g. a mixture of a Poisson or
         negative binomial distribution and a point mass at zero.

The computational method is based on the Laplace approximation
for integrating out the random effects, together with the
option of employing importance sampling at the posterior mode
of the random effects to permit arbitrarily close approximation
to the exact MLE. (However for these models differences appear to be
very small.)

Some of the generic convenience functions, such as
predict(), fitted.values(), ...  are still missing from
this package, but will hopefully be added in later
versions (contributions/suggestions are most welcome).
Other response distributions than negative binomial or
Poisson could easily be added.

Download site:

http://otter-rsch.com/admbre/examples/glmmadmb/glmmADMB.html

The package is based on the software ADMB-RE, but the full
unrestricted R-package is made freely available by Otter Research Ltd
and does not require ADMB-RE to run with user supplied data.

I you will find this useful,

Hans Skaug

-- 
Hans  Skaug

Department of Mathematics
University of Bergen, Norway
email: 	skaug at mi.uib.no
ph. (+47) 55 58 48 61