Message-ID: <53C2E25A-2182-45A7-9403-6036FCBF3DD4@mail.aub.edu>
Date: 2015-03-05T11:08:37Z
From: Zahwa Al Ayyash (Student)
Subject: Zero-inflated mixed models
In-Reply-To: <loom.20150304T031529-621@post.gmane.org>
Thank you both for your responses!
> On Mar 4, 2015, at 4:21 AM, "Ben Bolker" <bbolker at gmail.com> wrote:
>
> Zahwa Al Ayyash (Student <zsa11 at ...> writes:
>
>>
>> Dear all,
>>
>> Zero-inflated models have a count component
>> (Poisson/Neg. Binomial) and a zero component (logistic
>> regression part).
>
>> glmmADMB supports the zero-inflation feature but only through
>> estimating a zero-inflation parameter that is assumed constant
>> across the whole data set?.
>
>> 1. ?Why doesn't the glmmadmb support the two-stage fitting of
>> zero-inflated models like it does with hurdle models? I once read
>> that its estimation might be more complex than that of hurdle
>> models. Is that true? If yes, why?
>
> It doesn't support it because we never got around to implementing
> it -- i.e. making sure the interface was well designed, implementing,
> and testing.
>
>> 2. Is there an R package that allows for the two-model fitting
>> (i.e., each model with its own covariates) of ZIP or ZINB (not
>> hurdle) and allowing for a random effect?
>
> You can take a look the stuff here:
>
> https://groups.nceas.ucsb.edu/non-linear-modeling/projects/owls/WRITEUP/owls.pdf
> https://groups.nceas.ucsb.edu/non-linear-modeling/projects/owls
> https://groups.nceas.ucsb.edu/non-linear-modeling/projects/owls/R/owls_R_funs.R
>
> search for 'zipme.f' in the source code.
>
> Ben Bolker
>
> _______________________________________________
> R-sig-mixed-models at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models