Page MenuHomePhabricator

[SessionLength] Add a column to session tick data to capture sampling rate
Closed, DeclinedPublic

Description

While our first implementation of SessionLength uses the same sampling rate across wikis, we ultimately want to be able to vary sampling rates across wikis (and configure this over time, as traffic and computing power change).

For reporting and analysis purposes, we need to know the sampling rate so we can:

  • Estimate session counts, especially in automated reports
  • Appropriately interpret the accuracy of sampled data

To meet this need, we need some automatic way to capture the sampling rate as it applies to specific wikis and dates. One option for doing this is to send the sampling rate with every session tick event.

Event Timeline

LGoto triaged this task as Medium priority.Mar 24 2021, 6:12 PM
LGoto moved this task from To Do to Backlog on the Better Use Of Data board.
JVanderhoop-WMF subscribed.

This will be achieved by offering SessionLength as one of the standard metrics to use in experimentation T406261: [EPIC] SDS 2.4.3 Create a suite of standard metrics to use in experiments