Combined post for two minor releases from epidist and primarycensored respectively.
The main interesting updates to primarycensored is that the documentation was improved based on feedback from @kcharniga2 and there is a nice new attributes system from @pearsonca for better discovering analytical solutions when they exist.
The changes in epidist are more major with the addition of a new “marginal” model. This actually wraps the likelihood from primarycensored so if you were looking at that package and thinking I would like to fit more flexible models than this then now is your lucky day. There is now also more support for aggregate data types which can speed up model fitting but also make peoples lives nicer.
For me, this is quite exciting as its starting to show how the ecosystem can be joined up and how we can use low level insights (the analytical solutions) to really drive massive benefits for applied users (20 times or more speed up).
Both packages have planned future developments with primarycensored heading towards a write up in the near future led by @sambrand and potentially some more analytical solutions for common distribution pairs. epidist is still a bit more experimental so future releases are likely to overhaul how the simulation code is working, potentially do some refactoring of the non-tidybayes post processing and generally go over the documentation to make it easier for users to discover all the cool functionality. The write-up for all this stuff is still a bit away and I think at the moment its unclear who is going to lead (give me a ping if interested). I expect it to look something like @sangwoopark delay estimation paper with an extension to stratified partially pooled delays (but just comparing the naive, latent, and marginal models). I have also spoken a bit to @kylieainslie about whether we can include her cool work from mitey so I guess watch this space for that.