This PR closes #125. Instead of joining the epidemic phases by date, we instead left join to forecast date and location -- so any scores from a single forecast date and location are assigned a single epidemic phase corresponding to the phase assigned using the retrospective R(t) (so with future knowledge).
This PR closes #125. Instead of joining the epidemic phases by date, we instead left join to forecast date and location -- so any scores from a single forecast date and location are assigned a single epidemic phase corresponding to the phase assigned using the retrospective R(t) (so with future knowledge).
Revised figures here