Summary
Validator currently only works on daily data, which means that it cannot be organically ran on sources like nchs-mortality.
Possible Solutions
Releasing a version of the validator that only does checks unaffected by the weekly nature of the data (e.g. checking for missing date files, missing geo-type signal-type combinations) - will still need to tweak checks (e.g. making check_missing_dates only check for the first date of an epiweek)
Adapting parameters to consider the last N data-points (a.k.a weeks) rather than the last N days (Issues with this: Using data from 14 weeks ago as a reference may not work as well in determining things like outliers or trends)
qx-teo
commented
2 years ago
Summary Validator currently only works on daily data, which means that it cannot be organically ran on sources like nchs-mortality.
Possible Solutions