Retroactively correct anomalies in the data?

[mathiasbynens]

https://github.com/mathiasbynens/covid-19-vaccinations-germany#anomalies-in-the-data

We could look at https://impfdashboard.de/static/data/germany_vaccinations_timeseries_v2.tsv (nation-wide numbers only) and possibly figure out the correct numbers based on that (if we assume only the data for one state was wrong for any given date).

[mathiasbynens]

For 2021-01-25, Impfdashboard data lists dosen_erst_differenz_zum_vortag=58194. Cross-referencing this with the totals for this date + the day before, and assuming only Bayern overreported data for that date, this corresponds to firstDosesCumulative being overreported by 25876 and secondDosesCumulative being overreported by 14768.

This results in firstDosesCumulative=264807 and secondDosesCumulative=78860 for Bayern on 2021-01-25. Unfortunately, we don't know how this difference is split between firstDosesCumulativeBioNTech or firstDosesCumulativeModerna, so we can't adjust those values.