Thank you all
I got the point and after taking the map in consideration it seems quite reasonable and predictable.
Now I would like to shift discussion to the underlying essence of the question:
I'm going to use clusters to create local ratings (FYI only, nonofficial) using publicly available data @ WinSplits
Cluster is a good "entity" as club is usually too small to have sensible rating in itself.
Currently I have for last 12 months ~ 10K records out of which
(by hosting club)
2.4K is BAOC,GCO
300 is LAOC,SDO
800 is NEOC, CSU, UNO
800 is USMAOC,EMPO,WCOC,HVO
1200 CNYO,ROC,BFLO
1600 DVOa, COC, SVO
900 COC,SAMM
300 OCIN,MVOC,COO
400 LROC,RMOC
Most of the numbers would be greater if WinSplits were used more often. Apparently BAOC is much better than anybody in WinSplitting :)
What form of the cluster would be most optimal for the local B,C, meet low key rating?
P.S.
CNYO Cluster rating is here
http://tinyurl.com/nmlqxy
EMPO is here
http://tinyurl.com/nh82bw