We want to look at the R(t) estimate at the forecast date in each model compared to the forecast performance, so we can investigate differences in forecast performance as a function of the epidemic trend. Initial idea is to make a scatter plot of CRPS vs R(t) (either the median R(t), or could be the full distribution but I think that plot might look a bit odd). In order to do this, we need to save the R(t) estimates
Requirements
[ ] add an additional output to post_process that saves a single snapshot of the R(t) estimate at the forecast date
[ ] use this to make a plot of CRPS vs R(t) for each model
Open questions
Should we also save the R(t) quantiles for the calibration, nowcast, and forecast period?
Goal
We want to look at the R(t) estimate at the forecast date in each model compared to the forecast performance, so we can investigate differences in forecast performance as a function of the epidemic trend. Initial idea is to make a scatter plot of CRPS vs R(t) (either the median R(t), or could be the full distribution but I think that plot might look a bit odd). In order to do this, we need to save the R(t) estimates
Requirements
post_processthat saves a single snapshot of the R(t) estimate at the forecast dateOpen questions