Sorry for the delay
.
But what I meant was something like Mod A and Mod B had both given out 20 infractions, but Mod A had given 10 of those infractions to a single user, and the other 10 were to 10 other users, while Mod B gave 4 infractions each to 5 different users.
Attention, not really small:
As explanation: Every box plot is one mod. Each mod is "identified" with the number of infractions he/she handed out (there's a slight inaccuracy to the numbers of max. -1 to the numbers in the OP; no idea where that's coming from; does not include bots).
The boxes show in which area of # of infractions 50% of the mods infractions were...er...so...the mod to the left with 602 infractions had 50% interactions with users who received 1 - 3 infractions, whereas the following mods had to 50% only 1-2 infractions per user. The outlying arms represent 75%. The red bar is the median, which is nearly everywhere at 1, the red cross is the average, which is only in 3 cases higher than 2 infractions/user.
All the green circles identify users, which have received more infractions than 75% of all users from that mod (besides some of the left, which are below or direct at 75%; that's due to a bug in my script, sorry). The numbers at the plots show how many users, the size of the circles scales with that.
...okay...information overkill, I guess
.
One example, there's the mod with 11 infractions. The box and the arms are on the same size like the median. Means 50% of all users as well as 75% of all users received only a single infraction from that mod (and the median is in the middle of all infractions, which is then also 1). There's only 1 user with more than one infraction, with 3. The mod has in total 11 infractions, 3 to that single user, and obviously the rest of 8 for single other users. Makes 11 infractions for 9 users, which brings the average a bit higher than 1.
Or the mods to the right, the one with 4 infractions (ignore the 1s about the points): 50%, 75% as well as the median is at 1, the average is 1, no infractions outside of the 75% -> 4 users, everyone a single infraction.
What else can we see: More infractions -> more interactions with single users. Well, you'd expect this just by change, wouldn't you?
We can also see, that nobody in the group of top offenders with 20+ infractions received infractions by a single mod, because the max. infractions of a mod per user is at 15.
Other notable points: The user who received 6 infractions from the mod with 15 infractions in total (so the maximum point in the right half of the diagram) was a troll, who was permabanned later, so that's an exceptional case (and it's clearly visible that it is exceptional).
Else I'd say the distribution is relatively like expected, no really big outliers or peaks anywhere.
Not sure when I'll have the chance to look at the other data.
To calculate such things like in which area the most infractions were handed out, the number of reports within these infractions, etc. would require to download all 5000 infractions, + the infracted reports. With a waiting time of 3 seconds between every download and an estimated download time of 2 seconds, this makes 50 000 seconds, or nearly 14 hours...that's a bit much.
.
mmhh...I'll maybe see when there's the lowest server load to start a download.
.
).
