Clustering with constraints on centers and sizes

Viewed 11

My knowledge about clustering algorithms is quite low. I am looking for a solution of the following problem, the simpler the better, I need a reasonable solution easy to implement, not necessarily the state of art.

Given N points in 2D I need to find K clusters such that:

  1. each of the K cluster contains a fixed number K(i) of point and the sum for i = 1,...,K of all the K(i) point in the clusters is the total number of point N;
  2. each center C(i) of the clusters dominates or is dominated by the other centers in this sense: given the coordinates (x,y) of C(i) and (w,z) of C(j) then (x > w AND y > z) OR (x <= w AND y <= z).

Any idea?

0 Answers
Related