|
in reply to Re^2: Querying Saints in our Book (2)
in thread Querying Saints in our Book (2)
suggestions about whom to exclude or modify, as inactive users who get the overwhelming majority of their points via voting are not really representative of the ratio
I added fields in "Ranking the Saints by XP Efficiency" to require having visited the site in the most recent N weeks, and having made a minimum number of posts. Did not add a filter to require having posted in the last N weeks, which would have improved the quality of the rankings even more, but I think it could be done. Also my script can be configured to deduct 0.5 XP per day in an attempt to get closer to the XP gained from posting rather than longevity.
The way forward always starts with a minimal test.
|
Re^4: Querying Saints in our Book (2)
by chacham (Prior) on May 17, 2016 at 16:08 UTC
|
require having visited the site in the most recent N weeks
That helps give the recent average, which might be interesting, assuming the average has changed. But that's hard to detect, unless included when people first logged on or started posting.
having made a minimum number of posts
That's a decent idea. The question is, what is the minimum?
To get into SioB, a certain amount of reputation must be reached, that is, 3000. There is no minimum on posts.
MONK_LEVEL MONK_TOTAL MIN MAX AVG MEDIAN
+ LOWER EQUAL HIGHER
---------- ---------- ---------- ---------- ---------- ---------- ----
+------ ---------- ----------
13 205 0 903 238 229
+ 109 2 94
14 145 1 1009 328 312
+ 75 0 70
15 80 4 1236 456 445
+ 42 0 38
16 77 0 1133 553 538
+ 40 0 37
17 44 13 1582 751 767
+ 21 0 23
18 42 51 2565 936 859
+ 23 0 19
19 34 16 4183 1374 1347
+ 18 0 16
20 26 186 4500 2021 2057
+ 13 0 13
21 18 600 5335 3065 2974
+ 10 0 8
22 8 607 4520 3159 3650
+ 3 0 5
23 6 3416 7240 5503 5794
+ 2 0 4
24 2 3001 6420 4711 4711
+ 1 0 1
25 3 6323 7909 6956 6635
+ 2 0 1
28 4 10463 22669 17635 19772
+ 2 0 2
692 0 22669 785 378
+ 531 0 161
The average Writeups per level does seem to be mostly balanced. (Included the median, per LanX's suggestion.) How exactly would we determine the lower bar for Writeups?
deduct 0.5 XP per day in an attempt to get closer to the XP gained from posting rather than longevity.
That's an intriguing idea. On what is the half point based though? | [reply]
[d/l]
[select] |
|
|
"require having visited the site in the most recent N weeks"
That helps give the recent average, which might be interesting, assuming the average has changed. But that's hard to detect, unless included when people first logged on or started posting.
Hm, I was interested in the recent numbers, frankly, because I felt that the environment had changed so much from the early days of the monastery that it's almost not valid to compare a
monk who joined a year ago with one who was active from 2007-2008, say.
"having made a minimum number of posts"
That's a decent idea. The question is, what is the minimum?
Perhaps a function of the overall median reputation per post and the number of their posts? I.e. if your total XP is N and you haven't made at least M posts, your XP is deemed to be overweighted by longevity? Or perhaps better as a function of median rep per post and age as a monk - you should have at least this many posts after this long?
LOL, it's amusing. My goal was to try to evaluate contributions rather than just showing up ...
"deduct 0.5 XP per day in an attempt to get closer to the XP gained from posting rather than longevity."
That's an intriguing idea. On what is the half point based though?
From the Monastery FAQ:
Visiting the site regularly.
You have 25% chance of gaining 2 XP points once a day if you were logg
+ed in within the past 24 hours from when the vote fairy does the roun
+ds.
Cheers!
The way forward always starts with a minimal test.
| [reply]
[d/l] |
|
|
the environment had changed so much from the early days of the monastery that it's almost not valid to compare a monk who joined a year ago with one who was active from 2007-2008
That gets tricky though. The higher level monks are most likely to have been here a lot longer, so, if ancient history is deemed not valid, their ancient reputation should suffer the same fate.
Perhaps a function of the overall median reputation per post and the number of their posts? I.e. if your total XP is N and you haven't made at least M posts, your XP is deemed to be overweighted by longevity?
There are two issues with longevity. One, the login points you mentioned from the FAQ, and two, the rep granted from voting itself. The former can be negated somewhat with your idea of .5 per day, the latter cannot be detected and is would seem really hard to determine, except for the extreme cases (of almost no posts). Further, since everyone has a chance of getting the voting points, and the higher level monks have more points to give (and thus get more back) it would would be inaccurate to deduct them somehow from some monks but not others. So...
Or perhaps better as a function of median rep per post and age as a monk - you should have at least this many posts after this long?
I would think average, not median. Especially as the average does seem to be near the middle of most levels. With such wide ranges, i am afraid the median would be too far off.
Regardless, if we determine the average XP per post for that level (the average changes per level considerably, as can be seen above) we might be able to calculate the deviation or outliers for the individual monks. But how?
You have 25% chance of gaining 2 XP points once a day
Thanx. I had completely forgotten about that, or perhaps, just thought it only applied to the earlier levels. (That is, mixing it up with the special Initiate_bonus XP mentioned in Levels of Monks.) But, i still wonder, as not everyone logs in everyday. Some people only login during work days, others, only when they need something. Perhaps the regulars login every day, but that might be limited more to the higher levels, which would lower their average even further. And, the SioB "Last Here" column is not very granular with "months" and "years" masking a lot of the detail. Though, at least that much could be addressed by visiting the monk's actual node, which may cause undue stress on the server. So, i'm not sure that's a good idea, in this script, at least.
| [reply] |
|
|
Just released version 1.2.0 which includes Login_First, and Login_Last. (Unfortunately, the query is now even more Oracle specific, which was until now, iiuc, not so hard to change.) This allows querying age, though without much granularity. (The queries used are in the script. NodeReaper "never" first logged in, but last logged in 110 years ago. So, it gets a NULL.)
Years logged in:
YEARS AMOUNT
----- ----------
0 1
1 4
2 8
3 16
4 29
5 38
6 45
7 40
8 61
9 65
10 42
11 45
12 70
13 57
14 70
15 68
16 34
1
Anyway, a lot of the monks have been here quite some time. Just more food for though when working out efficiencies.
| [reply]
[d/l]
[select] |
|
|
