Thread: Seeding Calculations
View Single Post
  #14   Spotlight this post!  
Unread 23-06-2002, 21:56
archiver archiver is offline
Forum Archival System
#0047 (ChiefDelphi)
  
Join Date: May 2001
Location: Pontiac, MI
Posts: 21,214
archiver has a reputation beyond reputearchiver has a reputation beyond reputearchiver has a reputation beyond reputearchiver has a reputation beyond reputearchiver has a reputation beyond reputearchiver has a reputation beyond reputearchiver has a reputation beyond reputearchiver has a reputation beyond reputearchiver has a reputation beyond reputearchiver has a reputation beyond reputearchiver has a reputation beyond repute
scoring must be simple : ranking is an open question
Posted by Joe Johnson.   [PICTURE: SAME | NEW | HELP]


Engineer on team #47, Chief Delphi, from Pontiac Central High School and Delphi Automotive Systems.

Posted on 8/31/99 7:51 PM MST


In Reply to: Re: statistical bonus points posted by Jon on 8/31/99 10:05 AM MST:



I agree with you that scores must be simple to calculate at the end of a match.

This is not the same as saying that the ranking system has to be simple.

Because we have a limited sample size and because individual team performances are confounded by factors (not the least of which is the strength or weakness of you alliance partner(s)), I think that it is fair to try to get the best predictor of 'true rank' that we can.

I am not a statistical wizard, but I am concerned about finding a formula that we can all live with (and perhaps even explain in broad brush terms to our grandmas) while at the same time giving the best results we can get from our confounded, limited sample size.

Those who are statistical wizards notice that I did not propose some sort of performance index combined with a statistical correlation algorithm, which by the way I think could be an even better predictor of 'true rank' but I don't dare try to explain to my grandma ;-)

Joe J


__________________
This message was archived from an earlier forum system. Some information may have been left out. Start new discussion in the current forums, and refer back to these threads when necessary.