View Single Post
  #14   Spotlight this post!  
Unread 02-07-2007, 10:28
gblake's Avatar
gblake gblake is offline
6th Gear Developer; Mentor
AKA: Blake Ross
no team (6th Gear)
Team Role: Mentor
 
Join Date: May 2006
Rookie Year: 2006
Location: Virginia
Posts: 1,942
gblake has a reputation beyond reputegblake has a reputation beyond reputegblake has a reputation beyond reputegblake has a reputation beyond reputegblake has a reputation beyond reputegblake has a reputation beyond reputegblake has a reputation beyond reputegblake has a reputation beyond reputegblake has a reputation beyond reputegblake has a reputation beyond reputegblake has a reputation beyond repute
Re: [FVC]: Analysis Shows Improvement Possible in Ranking System

Quote:
Originally Posted by billw View Post
... Sean, your comments made me curious as to the probability of repeat partners (offensive or defensive) that a team would typically encounter when assigned on a random basis. Obviously the probability goes up as more matches are played and fewer unseen partners remain ...
Bill,

ManicMechanic (Yolande) and I have spent some time figuring out how to schedule matches that create no repeat opponents or allies for any/all team(s) until all the other teams in a tournament have been used in one of those capacities (i.e. until one is forced to start seeing repeat opponents/allies).

The work has gone reasonably well (and I confess that I haven't dug through any archives to see how many times this has already been solved in the past - I'm doing it for fun as much as to create a useful tool). When finished it will be a Java program that include a GUI for not only showing the match "pairings" but will show info to help scouts and to help teams focus on just their own matches, if they care to.

Success with this project should make it unnecessary to see repeat opponents/allies; and if the core algorithm is adopted by APs, it should improve the last-season method of running FVC Qual matches.

Side notes:
- I haven't tried to extend it to 6-team FRC matches yet; but I am optimistic that I will be able to do that.
- So far, the more general of the two scheduling algorithms we are working on is not a closed-form equation or procedure that is guaranteed to work in all circumstances. At the present it is a set of heuristics (That work well but aren't perfect).

Blake
PS: Does anyone out there want to do an archive search for me; or does anyone have info from other sources about past or current algorithms?
__________________
Blake Ross, For emailing me, in the verizon.net domain, I am blake
VRC Team Mentor, FTC volunteer, 5th Gear Developer, Husband, Father, Triangle Fraternity Alumnus (ky 76), U Ky BSEE, Tau Beta Pi, Eta Kappa Nu, Kentucky Colonel
Words/phrases I avoid: basis, mitigate, leveraging, transitioning, impact (instead of affect/effect), facilitate, programmatic, problematic, issue (instead of problem), latency (instead of delay), dependency (instead of prerequisite), connectivity, usage & utilize (instead of use), downed, functionality, functional, power on, descore, alumni (instead of alumnus/alumna), the enterprise, methodology, nomenclature, form factor (instead of size or shape), competency, modality, provided(with), provision(ing), irregardless/irrespective, signage, colorized, pulsating, ideate
Reply With Quote