Skip to content
Prev 9137 / 29559 Next

splitting a map into areas by number of addresses

On Aug 26, 2010, at 10:23 AM, Barry Rowlingson wrote:

            
Graph partitioning is NP-complete. However, heuristics exist that work  
in a resonable time for real world graphs. see http://en.wikipedia.org/wiki/Kernighan%E2%80%93Lin_algorithm 
. But I doubt if this would work for your situation.
Another idea, is to create 1000 clusters (not sure how you would  
ensure exactly 1000 points in each).  Find the centeroid of those  
clusters and then create the Voronoi tessellation.