Description
Add support for the geometric distribution and the corresponding GLM. It's a special case of the negative binomial distribution (when n=1 in the usual parametrization) which is already supported, so I believe it should be relatively straightforward to implement. Arguably, geometric distribution is not some exotic distribution, so users may expect to have that available.
The expected advantages of having this are:
- Prettier/better readable stan code
- Possibly some performance gain compared to using the NB distribution
Description
Add support for the geometric distribution and the corresponding GLM. It's a special case of the negative binomial distribution (when n=1 in the usual parametrization) which is already supported, so I believe it should be relatively straightforward to implement. Arguably, geometric distribution is not some exotic distribution, so users may expect to have that available.
The expected advantages of having this are: