Again, a fascinating thread. And amazingly I actually studied this kind of problem in university 20 years ago (shudder) under a brilliant professor called Ed Jernigan (University of Waterloo, faculty of Engineering, department of Systems Design).

So for scholarly research I can suggest Google.

Abigail-II's point is quite valid -- there need to be constraints on the solution otherwise optimization cannot be done. You need to do something like

• set a maximum cluster radius and
• set a minimum and/or a maxmimum number of complaints within each cluster.

--t. alex

"Mud, mud, glorious mud. Nothing quite like it for cooling the blood!"
--Michael Flanders and Donald Swann