stephenbrooks.orgForumOtherBrowser WarsHow do the statistics work?
 Alex2004-09-16 04:21:17 Hi there!I'd like to know how the statistics on the game page are calculated.  I can't understand the following thing:Firefox has 1000 wins and K-Meleon 1350 - but if K-Meleon wins, his percentage of won games rises much more than the percentage of firefox if he won a game - how can that be? Stephen Brooks2004-09-17 07:24:02 Well, it says in small type at the bottom of the page that the figure is games won over games "participated in". I think you count as participating in a game if you place 3 or more pieces (though it might be 4).  Anyway, it could be that FireFox has been spreading itself rather thin and playing several counters in games where, say, Opera and IE finished it off between them.
 Alex2004-09-18 14:46:11 I thought, the number of games displayed left of the percentages is the number of games won.If that's true, Firefox had participated in about 2300 games (with 43-44% and 1000 games won) and K-Meleon in about 2900 games (with 46% and 1350 games won) - so one in 2900 games should make a smaller difference in the percentage than one in 2300 games - you know what I mean?Well, maybe it would be a good idea to show some more information like the complete number of games participated in etc - also some more statistics on the page with the combination listings would be great.Nevertheless, this war is the best fun on the internet at the moment - and it's a real problem when playing it at work
 Alex2004-09-23 06:36:43 I observed the stats now for some games and found out the following:K-Meleon gains about 0,3% each game he won.1386 games won with 50% = K-Meleon participated in 2772 games.Some minutes later:1391 games won with 51,94% = K-Meleon participated in 2678 games...If it was the other way around, it would maybe make sense - but not this way...Would it be a big effort to include the number of games participated in on the game page?  You could show a little table with the games in total, the games won and the percentage that's resulting from the two numbers before... - that way the statistics would be much more understandable! Stephen Brooks2004-09-24 02:18:31 You're right, that does seem strange.  Though there's been something a bit strange with the stats since the "complicated ones" (individual breakdown of game combinations) got reset at one point.  It might be those percentages are actually being calculated out of a much smaller pool than the totals (which haven't ever been reset). Alex2004-09-24 02:57:51 Yeah, that sounds logical - if K-Meleon was registered with it's own icon long before this reset and firefox appeared afterwards, it could be that the percentages of firefox are more realistic than the stats of K-Meleon.Do you have the "real" numbers of games played in total anywhere in the database?  Maybe it would be possible to correct the stats with them... - or you could adapt the total stats to the "complicated" ones... Stephen Brooks2004-09-29 14:25:48 Now if you mouse-over the percentages it will display the fraction that it was calculated from.  That fraction comes from the 'advanced' stats showing the breakdown of individual game combinations, which got reset at some point.
 Alex2004-09-29 23:34:26 Big thx man!That way the stats make sense!  Well, I'd put these numbers directly behind the percentages or give a hint in the description, so that everyone knows of it and not only the two of us ...Now I understand, how those huge differences occur...
 Alex2004-09-29 23:36:23 sorry for this triple post - but the message window said "the database is not available" after posting the message - and the second time it said, that the same message had been posted before...
E-mail: sbstephenbrooks.org
 Yahoo: scrutney_mallard Jabber: stephenbrooksjabber.org