Scenario: I have a list of addresses (they have already been converted to lat/long) and the conference that the person attended.
What I want: I want to add a circle shape to a Virtual Earth map that covers, let's say, 90% of the attendees to a conference. I will repeat this for each conference. This will show me how much the different conference "attendance areas" overlap each other. The reason for the 90& is to exclude the "outliers", those that aren't in the main "attendance area".
So the question: What is the algorithm or SQL spatial query do get the above data set (the 90%)? I think I can take it from there (ie. how big to make the circle, where to place it, etc.)
For each person, total the distances to all other people. Sort by these totals, and remove the highest 10%.
Then take to two remaining people that are farthest apart, and place your center there. The distance between the farthest people is your diameter.