This means we will not be able to do any forecasts of any length since we do not sample for these parameters beyond the dates of the case data, I believe this is not intended, but suggest we discuss first, in case I missed if we decided to include forecast implementation elsewhere.
The fix here is easy but suggest we do it after merging other changes
AugustHao
commented
2 months ago
create_observation_model outputs objects size and prob, used to realise neg binom draws of cases, but note the current date lengths of these objects are based on observed case dates (see https://github.com/idem-lab/lowerGPreff/blob/0aa5453de2af075afaa2ba6c4b2474a4aa88a29f/R/create_observation_model.R#L97 and https://github.com/idem-lab/lowerGPreff/blob/0aa5453de2af075afaa2ba6c4b2474a4aa88a29f/R/create_observation_model.R#L85)
This means we will not be able to do any forecasts of any length since we do not sample for these parameters beyond the dates of the case data, I believe this is not intended, but suggest we discuss first, in case I missed if we decided to include forecast implementation elsewhere.
The fix here is easy but suggest we do it after merging other changes
For reference see the old code where these objects are as long as the infection dates, which include the extra forecast days https://github.com/AugustHao/convGPreff/blob/db1b82a359fa5a4592515a3b3cce8af4b924e460/R/reff_model.R#L100