Page MenuHomePhabricator

Approximate mediawiki-history user creation dates using user-id/registration-date coherence
Open, LowPublic

Description

Since https://gerrit.wikimedia.org/r/c/analytics/refinery/source/+/454338, we leave users events without registration date nor edit activity out of our dataset. An nice way to keep them would be to approximate the creation date using other users having creation date set and being close in term of userIds:
The registration date we could use could either the one of the user having the biggest userId and having a registration date and whose Id is smaller than the one of the user we want to extrapolate. We could also do it with the smallest userId whose bigger than the one of the no-registration-date user.