-1

What is the best method to cluster points of a set with a maximum diameter as threshold and as few clusters as possible?

So for example having a set of gps points and a diameter threshold of 500m. How to use an algorithm to minimize the count of clusters?

As I am using python, is there any package yet?

Timo
  • 21
  • 5
  • I already tried the dbscann algorithm but in my understanding, the algorithm uses the epsilon value only between to points and not consider the diameter. How can I ensure not having a greater diameter? – Timo Dec 08 '17 at 11:08

1 Answers1

0

Use complete linkage clustering. I am surprised you didn't know this method, it is one of the oldest. Here also is Leader, but that one is really low quality.

You cannot afford to check all possibilities (which would be NP hard), but there are algorithms to find a good solution in O(n²) and O(n³).

Has QUIT--Anony-Mousse
  • 76,138
  • 12
  • 138
  • 194