good chemistry is complicated, and a little bit messy -LW |
|
PerlMonks |
Re: Efficient Assignment of Many People To Many Locations?by kvale (Monsignor) |
on Feb 25, 2005 at 09:58 UTC ( [id://434417]=note: print w/replies, xml ) | Need Help?? |
I don't know if this is a problem in AI, but it certainly is a problem in operations research.
Off the top of my head, there are a few approaches you could take. Your stochastic search method seems reasonable and I would add to it that you should run it several times with differing initial conditions. Another approach would be a genetic algorithm with a permutation-based chromosome and crossover scheme (Goldberg talks about these methods in his book on genetic algorithms). Often times, simple heuristics can get you a pretty good solution in a minimal amount of time. I think a greedy heurisitc would work well here. The idea is to create the N*M matrix of distances between N salespeople and M locations. Then find the smallest element in the matrix. At the smallest element, assign that salesperson to that location and delete the corresponding row and column. Then find the smallest element in the reduced matrix and make another assignment, etc. Continue until all salespeople are assigned. -Mark
In Section
Seekers of Perl Wisdom
|
|