- DAPPER is not made to assimilate real data cases.
- It requires a truth time series for its statistics and liveplotting.
- Corona1 model is ridiculously oversimplified, using no compartments.
- It is not chaotic, but highly sensitive to the R parameter,
which is modulated by societal intervention.
- The biggest question seems to be the modelling of the observation operator.
- How to deal with numbers counted on different days.
- How to deal with test volume vs number of positives.
Cross-validate with "future".
Example should say: