Research talk:Anonymous editor acquisition/Signup CTA experiment/Work log/2014-06-10

Tuesday, June 10th

Not much time today. I'm hoping that I can get the basic bits of an overall productivity analysis together before I call it quits today.

So, I want to compare the overall productivity of all anons who entered the funnel during the experimental period. This means that I need to measure edits performed by these users whether they registered or not. --Halfak (WMF) (talk) 21:58, 10 June 2014 (UTC)

---

My most basic measure of productivity is a count of the number of edits performed. My plan is to build a log normal test based on the count of edits saved during the experimental period.

SELECT
    token_info.wiki,
    token,
    SUM(event_revId IS NOT NULL) AS experimental_revisions
FROM staging.token_info
LEFT JOIN log.TrackedPageContentSaveComplete_8535426 tr ON
    token_info.wiki = tr.wiki AND
    token_info.wiki = tr.event_token
WHERE timestamp BETWEEN "20140519180800" AND "20140526180800";

Sadly, the event_token column on TrackedPageContentSaveComplete_8535426 isn't indexed, so this'll take a while. --Halfak (WMF) (talk)

---

Too long. I'm changing the query so that I only have to do this once. I'll do my aggregation in another pass.

SELECT
    wiki,
    event_token as token,
    event_revId as rev_id
FROM log.TrackedPageContentSaveComplete_8535426
UNION
SELECT
    wiki,
    event_token as token,
    event_revId as rev_id
FROM log.TrackedPageContentSaveComplete_7872558;

This will give me much more flexibility anyway. --Halfak (WMF) (talk) 22:20, 10 June 2014 (UTC)