Another distribution

[Moisés Augusto]

Hi, I want to incorporate another probability distribution to the counts than the ones offered in the “family” argument from INLA function. 

Can I adapt one of the options (such the Poisson) to a desired distribution? 

[Helpdesk (Haavard Rue)]

Hi,

this has to be done directly in the source code and there is no
interface for likelihood as there is for latent models
(rgeneric/cgeneric)

If you send me details I can assist you more

Best
H

> --
> You received this message because you are subscribed to the Google
> Groups "R-inla discussion group" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to r-inla-discussion...@googlegroups.com.
> To view this discussion on the web, visit
> https://groups.google.com/d/msgid/r-inla-discussion-group/a2dd82d0-2281-4e80-80c6-d312841a9ab5n%40googlegroups.com
> .

--
Håvard Rue
he...@r-inla.org

[Moisés Augusto]

Hi, to be short, I want to incorporate the Bell distribution and use it to fit a spatio-temporal model. 

I tried to look on the INLA function code, but I didn't manage to find the family specification.

I was planning to find it and look to the specification of some distribution (such Poisson) and replicate this specification, but adapting it to the Bell. 

Do you have any suggestion? 

[Moisés Augusto]

Hi, how are you?

Sorry for bring this theme back.

I want to incorporate the inflated zero bell distribution.

There is a way I can code and incorporate it to INLA?

In the another time, I didn't find it.

The distribution is like the zeroinflatedpoisson1 (specified here: https://inla.r-inla-download.org/r-inla.org/doc/likelihood/zeroinflated.pdf)

The bell distribution, which is currently implemented, it's like in the pdf.

[Helpdesk (Haavard Rue)]

In short, likelihood models needs to be added directed into the source
code, there is no user-interface in R or C.

Let me check it, as most of the work was done already, the 'bell' one is
pretty involved.