Hi @sangwoopark,
Forecasting sounds like a cool idea but I’m worried that it’s going to depend heavily on priors
This is definitely the case though where the generative model is the same as before the forecast window those priors will at least have hyper priors informed by data (i.e in the case of a random walk). There are other specifications we could imagine (and would be supported via the formula interface) that make more sense. For example, if the temperature has a role and forecasts for that are available.
Also agree we should have an option to allow fixing R_t as that is so often what people/want expect to see and also can perform quite well given the uncertainties of doing anything else. I think this is what @Gunnar means when he says fixing the current dynamics.
The current code base can definitely be hard to navigate which is a shame. It might be worth opening a discussion to think about how we can improve this and in general make it easier to onboard those interested in contributing.