User talk:Amgine/Wikidata random sample test

Latest comment: 10 years ago by Hym411 in topic Other data

Other data edit

This is not the best methodology, and toolserver was giving me issues with getting data. The use of the tools in question is very low, and only the most active of a certain kind of user likely to use it. I went through RecentChanges (and then eventually only to RecentChanges on the user space). I added their username to this and identified the wikiproject they had the most edits on. After that, I went to wikidata edits by month, and recorded that. On the project they had the most edits on, I looked at edits by month. Real struggle to find users, so only 6 users in my sample... --LauraHale (talk) 20:30, 16 November 2013 (UTC)Reply

  Dcirovic Ladsgroup LauraHale SPQRobin Samat Hym411
  wikidata sr.wikipedia wikidata fa.wikipedia wikidata en.wikipedia wikidata nl.wikipedia wikidata hu.wikipedia wikidata commons
2012/10 49 600
2012/11 4 302 95 276
2012/12 0 135 11 269
2013/01 4 273 293 425
2013/02 317 6850 90 121 1264 110
2013/03 2166 11633 498 329 3 87 1710 200
2013/04 1560 7498 271 389 3 277 711 109
2013/05 1748 15669 206 161 70 74 117 76 54 16
2013/06 838 23981 114 324 48 69 89 56 381 134
2013/07 13457 114558 112 647 31 703 474 10 682 246
2013/08 475 529954 55 200 9 631 220 20 326 497
2013/09 232 203732 187 659 8 114 1494 36 497 748
2013/10 298 12598 112 254 12 32 1886 32 477 1689
2013/11 278 1829 159 125 119 140 1447 20 5 19 1673 1631
Correlation
0.211523956
#¡DIV/0! 0.6462335
Slope
0.161100441
#¡DIV/0! 0.48197443
Average difference 90693.3 162.0769231 202.6666667
14 124.4285714
Even with borders I'm not at all certain I understand what is being measured here. Amongst the sample you have here include months with more than 500,000 edits, and thus exemplify why trying to measure botter contributions is difficult - that's a bit more than 1 edit every 5 seconds 24/24h 30/30d. But if the tool does get me the frequencies then it will be much stronger than the simple 'snapshot' method I'm currently using.
I wanted to directly compare total volume of edits per month since a person started contributing to Wikidata with the volume of edits they make to their home wiki project. (Then calculate correlation, slope and difference in edit totals on average per month.) Goal to see if there is a correlation between contributions between Wikidata and other projects. --LauraHale (talk) 06:25, 17 November 2013 (UTC)Reply
<nods> Yep. The stats aren't reliable, and the measure is also suspect (bot or script based editing is as like hand editing as periwinkles are like periwinkles - not at all.) Excellent tools and study models. Thanks! - Amgine | t 06:34, 17 November 2013 (UTC)Reply
This method is likely to only pick up a certain type of user, the more power user. If doing an extensive study on this topic, I think you would really need about 4 different ways of approaching the data to try understand different user groups that interact with projects in their own ways. --LauraHale (talk) 06:40, 17 November 2013 (UTC)Reply
(Well, I have most edits on Commons and Wikidata, however my home wiki is ko.wikipedia. --레비ReviSUL Info 07:04, 17 November 2013 (UTC))Reply
Yes, I had already noticed this Hym411, and I am writing up a methodology to track wikidata, home wiki (where available), and largest contribution totals. One problem with this is some user accounts are not Global accounts, and therefore the 'home wiki' is not known. (for others, home wiki and largest contributions are the same and I have to figure out how to note this in the tabular output.) - Amgine | t 07:10, 17 November 2013 (UTC)Reply
You may use ref tag noticing it. --레비ReviSUL Info 07:17, 17 November 2013 (UTC)Reply
Return to the user page of "Amgine/Wikidata random sample test".