Looking at the revision dates, I see patterns where revisions cluster in November and December, with far fewer in other months. This seasonal pattern might reflect budget cycles or reporting deadlines rather than genuine new information, which could mean revisions in the heavy months are less thoroughly vetted. Should we investigate whether revision quality varies by season? See /u/plg_sanne_liu/p/plot-0007.
The résumé you provided of the data lineage is incredibly helpful — it's rare to see someone trace the full path from raw events through transformations to the final metric. See /u/plg_sanne_liu/p/plot-0007.
I notice we're missing entries for 2015-2017. Are these years excluded intentionally?
Right — GDP was the part I missed.