We discussed in the last meeting that it would be desirable to have a common document in which our model vision is collected and written down.
In my view, this document serves two purposes:
-
Communication of our vision to the outside world.
-
An internal basis to understand details and modelling approaches, for example when reviewing code of others or thinking about model specification or extensions.
A few thoughts on this:
-
Maybe both purposes need different levels of detail
-
Especially for the latter, I think it would be really helpful if the “small” details like e.g. clear definition of the range of indices, etc. are listed correctly.
-
It might make sense to create a static version of the document (with a focus on the outside communcation) and a more detailed document that evolves over time (not sure whether this works well though)
- It might also be nice to have some additional documentation how the mathematical description of the model is then finally implemented in STAN-code, but its probably also possible to cover much of this in the code documentation/comments.
Sam has offered to create a draft version of the document, but I thought it might be good to collect some material documenting aspects of the model already in this thread.
We can also use parts of the next meeting to discuss on the overall document/structure, missing parts, and who is willing to contribute at the next meeting!
2 Likes
Thanks @FelixGuenther,
The current write-up of the model is Model definition and implementation • epinowcast. There is a lot of context missing, some of which is provided in the referenced papers (i.e Felix’s and others). This write-up will be out of date when we add renewal processes, and latent reporting (i.e Flexible expectation model by seabbs · Pull Request #152 · epinowcast/epinowcast · GitHub). There will also be more context when @adrianlison makes his PhD work public as that serves as the basis for some of our new additions.
I agree about the split purposes. My proposal would be to write a mission statement with an overview of the model and our general aims as a community and then uses parts of this for package documentation (i.e perhaps for the www.epinowcast.org site).
I definitely agree we need more detail on code implementation as can differ quite a bit from the maths as written down. This could happen in a wiki like doc for high level details perhaps?
1 Like
Dropped the ball on this a little (as ever) but have now got a model outline in place to capture where we currently are (at least roughly).
As @FelixGuenther flags I think we also need a communication document for our long-term vision. I see that as being in the form of a short paper. I am planning on starting on this over the next few weeks. I’ll share the ongoing draft here so people can weigh in and we can discuss in more detail.