Juuni OFFTOPIC

[feelitwell]

fascinating kuidas kasutajad töötavad data ja numbrite kallal

[muttaja]

fascinating kuidas kasutajad töötavad data ja numbrite kallal

Jep, ise ka tunnustaks nende tööd. Lahe visualiseering:

http://http://forumserver.twoplustwo...&postcount=123

The "scientific" way of checking how different 2 accounts are from each other is to measure the distance between their respective stats. I did it for 1 banned bot accounts: BOOTTLEGGER, 3 active accounts from my list: IMissFender,Samanta81,SusaaNIN and 4 legit regulars from PL200: N@T@L@T,Shuller_A1t,Nakamator,Bagrovui. I picked those legit accounts because they have played a lot of hands and are mostly Russian.

The stats of banned bots/alleged bots are here http://i57.tinypic.com/2n84qow.png
I entered them into an excel spreadsheet here:


I will not post the stats of legit players to protect their privacy.

Afterwards I have calculated the distance between those 8 accounts mentioned earlier. I have used "Squared Euclidean distance" for it. It can be found here http://i61.tinypic.com/6ell61.png


Thats the distance between following account pairs in form of a graph



The 6 first columns are BOOTLEGGER,Samanta81,IMissFender,SusaaNiN compared in pairs.

The next 6 columns are 4 legit accounts compared in pairs.

The last column is an alleged bot SuSaaNiN compared to legit Russian grinder N@T@L@T.

Small column means small distance, accounts are very similar, big column means the accounts are completely different.

EDIT: I didnt enter more accounts because typing all of it into excel is a huge pain in the ass. I will put all my research on Google Drive on Tuesday and make it free for everyone to download, including the spreadsheet to calculate the difference between any 2 accounts.

[muttaja]

... kuigi mulle tundub, et hetkel näidatakse vaid seda infot, mida näha soovitakse. Võiks siis juba kõigi volüümi poolest top20 vms omavahelise antud koefitsendi välja arvutada, ja siis vaadata, kui tõenäoline on, et süüaluste antud näitajad niivõrd hälbivad.