Added functionality for joint modeling in rstanarm

I have locally implemented additional functionality to the joint modeling code in the rstanarm package in order to appropriately model negative binomial data in the longitudinal component as part of my doctoral thesis work. I am trying to determine whether it would be better to incorporate these changes into the existing rstanarm package or to create my own R package. I have made the following changes:

  1. Enabled offset values in the longitudinal component
  2. Introduced a scaling parameter for the association term in the survival component
  3. Implemented the muslope association structure

Would these changes would be appropriate for the rstanarm package?

Thanks!

Yes

Thanks so much for reaching out. We’re always happy to work with new developers and try to prioritize helping them make their first contribution.

The thing to do is create an issue in the RStan repo outling the change, then if people are OK with the design, create a pull request.

@pamelanluna In this case it would be the rstanarm repo, but otherwise what Bob said. We’re definitely happy to get contributions!