Update Account Deletion Rate to match data.login.gov#9516
Conversation
- Uses 'fully registered' users - Matches timestamp logic changelog: Internal, Reporting, Update monthly key metrics report
| def deletion_rate | ||
| deleted_user_count.to_f / users_and_deleted_for_period.to_f | ||
| deleted_user_count.to_f / fully_registered_users.to_f | ||
| end |
There was a problem hiding this comment.
One thing to note, I'm not 100% sure I agree with the calculation on data.login.gov, because "fully registered users" is not safe for backdating, the registrations logs rows are deleted and not recoverable, where the "total registered" is, we can approximate counts far back in time based on deleted_users
There was a problem hiding this comment.
For what it's worth, I'm not attached to that logic (as the person who wrote it), if we want to update to something more accurate and backport to data.login.gov .
But also we're tentatively considering to want to change this to an "Account Reset Rate" anyways, so maybe it's best to hold off for that.
| def deletion_rate | ||
| deleted_user_count.to_f / users_and_deleted_for_period.to_f | ||
| deleted_user_count.to_f / fully_registered_users.to_f | ||
| end |
There was a problem hiding this comment.
For what it's worth, I'm not attached to that logic (as the person who wrote it), if we want to update to something more accurate and backport to data.login.gov .
But also we're tentatively considering to want to change this to an "Account Reset Rate" anyways, so maybe it's best to hold off for that.
2 participants
🛠 Summary of changes
Updates the emailed Monthly Key Metrics report to use the same "deletion rate" math as data.login.gov
registered_atfor fully registered (source link)deleted_atfor deleted users (source link)