OK, gang, I erred, I goofed, I blundered, I . . . (dare I say it) . . . fumbled . . .
I know, I know, even Todd couldn't believe it (ok, well it's early Saturday morning so I haven't told him yet, but I'm sure he won't believe it when he finds out).
At any rate, I've literally spent hours trying to figure out why the initial run of the program kept placing Thomas County Central in with Alcovy and (actually more curious) placing Jefferson and Jackson County into different regions. I've combed through code, fiddled with parameters, reran numerous trials, lost more hair . . .
And suddenly there it was.
But first, a little background – The program is actually one I dusted off from the GHSA’s 2013 region realignment. Back then, along with the regular realignment scenario, I had also been trying a little experiment to not only find the shortest distances between teams to make regions but to match their enrollments as closely as possible as well. The program actively attempted to minimize the differences in the distance and the enrollment at the same time.
Don’t worry -- I completely remembered to remove the enrollment portion from that equation.
However, part of that same experiment was to try to match the overall size and average school enrollment of all regions in the same class. That part . . . well, I forgot about.
In the initial run from Friday, the average enrollment for the projected Class AAAAAA ranged from 1725 to 1864, a pretty darn tight spread for 60 teams placed into 8 regions, so I consider the program an overwhelmingly huge success!
Except that's not the goal here . . .
Anyway, I’ve corrected the formula and rerun the program and hopefully you’ll see improved results. The only other change I’ve made is a hard requirement that five schools be the minimum region size and three teams over the average region size be the maximum.
Also, in the interest of time I only ran 100 restarts, which is still plenty sufficient for waaaaaaay too early projections.
Note that this doesn’t take into account the isolation rule, but I’ve added some additional columns concerning enrollment size and distances that will help us see which teams might be good candidates for that. It also doesn't take into account teams that might opt to play up.
Finally, as a reminder, the “Group” designators are not intended to be the region designators, but are simply the numbering of each group going from south to north.
[CORRECT] Region projections (but still waaaaaaay too early . . .)
AvgEnrDiff = Average Enrollment Difference with other projected region members
AvgLnrDif = Average Linear Distance Difference with other projected region members
AvgDrvDif = Average Estimated Driving Distance Difference with other projected region members (AvgLnrDiff x 1.6)
All distances are one-way
