Page MenuHomePhabricator

Gather data about number of praise-worthy mentees
Closed, ResolvedPublic

Description

Within Personalized praise, we will recommend mentees to send praising messages to mentees that are doing well. In the first iteration, we want to do that by suggesting mentors praise mentees that meet all the following conditions:

  • have more than N edits in the last H hours,
  • have zero reverted edits

According to T322446, mentors will be able to customize the values N and H. We still, however, need to provide a reasonable set of defaults for mentors to start with.

To avoid guessing the right values of N and H, we should see how many praise-worthy mentees would be suggested per mentor (on average) for the following configurations:

  • exactly 1 unreverted edits in (timeframe)
  • 2-9 unreverted edits in (timeframe)
  • 10-19 unreverted edits in (timeframe)
  • 20-29 unreverted edits in (timeframe)
  • 30-39 unreverted edits in (timeframe)
  • 40-49 unreverted edits in (timeframe)
  • 50 or more unreverted edits in (timeframe)

where (timeframe) is 48 hours, 1 week, 2 weeks and 1 month.

The data should be gathered for the following wikis (pilots + two big wikis):

  • arwiki,
  • bnwiki,
  • cswiki,
  • enwiki,
  • eswiki,
  • frwiki

Event Timeline

It might be nice to also see data for variations of "N" (Number of edits), but if that's too complex, I'm fine with starting with this. Thanks!

I might be nice to also see data for variations of "N" (Number of edits), but if that's too complex, I'm fine with starting with this. Thanks!

Good point. That should be reasonable easy to do. I'd like to cover as many different Ns as possible, without also generating a high amount of data (to ensure we can go through the results).

Perhaps as a first step, we can bucket the numbers together (and in the future, we can look closely at any of the buckets). In other words, let's see how many users have:

  • exactly 1 unreverted edits in (timeframe)
  • 2-9 unreverted edits in (timeframe)
  • 10-19 unreverted edits in (timeframe)
  • 20-29 unreverted edits in (timeframe)
  • 30-39 unreverted edits in (timeframe)
  • 40-49 unreverted edits in (timeframe)
  • 50 or more unreverted edits in (timeframe)

and use timeframes from the task description/mockups. We'd then have 6 tables (one per wiki) full of numbers to analyze.

Does that seem good to you, @KStoller-WMF?

Urbanecm_WMF changed the task status from Open to In Progress.Nov 16 2022, 2:22 PM
Urbanecm_WMF triaged this task as High priority.

Most of the wikis defined in the description are below. Each number is an average number of praise-worthy mentees per mentor (from 2022-01-01 to 2022-09-31).

arwiki
time_bucket1-1 edits2-9 edits10-19 edits30-39 edits50-∞ edits
48h_id9.1246077991931877.2279411764705881.383707201889021.03883495145631071.02
day_id5.633266027258964.4081258191349941.17182497331910351.01.0
doubleweek_id36.7652370203160330.178082191780823.951.29629629629629631.280373831775701
month_id68.3556.568807339449547.0066225165562911.72941176470588242.0555555555555554
week_id22.30268199233716318.1024643320363162.50549450549450551.15044247787610621.1368421052631579
bnwiki
time_bucket1-1 edits2-9 edits10-19 edits30-39 edits50-∞ edits
48h_id2.32856066314996252.311403508771931.13559322033898311.01.0
day_id1.69960079840319361.73671013039117361.0656251.01.0
doubleweek_id8.1935483870967747.663580246913581.67647058823529421.05263157894736841.18
month_id14.84076433121019214.452.4141414141414141.21052631578947371.4444444444444444
week_id5.04.8282647584973161.3715596330275231.04347826086956521.0606060606060606
cswiki
time_bucket1-1 edits2-9 edits10-19 edits30-39 edits50-∞ edits
48h_id2.5034177724165662.3984803714647531.11652542372881361.01.0
day_id1.81519507186858321.72001121076233181.07280513918629561.02325581395348841.0
doubleweek_id8.622176591375778.929313929313931.91666666666666671.07575757575757571.0862068965517242
month_id16.0645161290322617.090909090909093.0946745562130181.22413793103448271.28125
week_id5.45027932960893855.5292096219931271.43882978723404251.04838709677419351.0196078431372548
eswiki
time_bucket1-1 edits2-9 edits10-19 edits30-39 edits50-∞ edits
48h_id6.8183856502242157.54141189549749851.73417721518987331.0639534883720931.0535714285714286
day_id4.3926417370325694.52888356779407351.32173174872665531.02721088435374151.0555555555555556
doubleweek_id23.22591362126245730.0363036303630375.8812260536398471.73103448275862061.8741721854304636
month_id37.4383561643835650.24489795918367510.52.67368421052631573.701923076923077
week_id15.11363636363636318.327586206896553.5118790496760261.37931034482758631.3121019108280254
frwiki
time_bucket1-1 edits2-9 edits10-19 edits30-39 edits50-∞ edits
48h_id5.5749471327279515.4403088998607421.4311844077961021.0366132723112131.017921146953405
day_id3.56166304424034233.4273050933662721.20698445067550341.0101265822784811.0
doubleweek_id20.9121887287024920.8230318802862733.95956873315363871.40963855421686751.4412371134020618
month_id38.35744089012517638.45254470426416.9736842105263162.13157894736842122.3263157894736843
week_id13.08333333333333412.7971326164874542.55592654424040071.21936758893280641.1513859275053304

Interesting, thank you so much for pulling this data! I have opinions... but I'll keep them to myself for now and let's chat with Ambassadors on Tuesday. I've added the topic to our agenda: Setting a default “praiseworthy mentee” baseline based on data.

Interesting, thank you so much for pulling this data! I have opinions... but I'll keep them to myself for now and let's chat with Ambassadors on Tuesday. I've added the topic to our agenda: Setting a default “praiseworthy mentee” baseline based on data.

Sounds great!

FTR, enwiki was originally in the specifications, but it turns out enwiki's too huge to be worked with in the same way as what I did with other wikis. Leaning towards keeping it out for now, and deciding only based on the wikis already listed in my previous comments.

Moving to Watching, as this is done. Keeping it opened for a while so we can discuss the data.

Urbanecm_WMF changed the task status from In Progress to Open.Dec 1 2022, 10:58 PM

Discussed with Ambassadors. I believe there was consensus that we should allow defaults to be set in community configuration, but that we can use this data to determine a decent baseline that communities can configure.