The goal of this method is to partition the points into
groups such that the sum of squares from points to the assigned
cluster centers is minimised.
The method works as follows [2]:
![]() |
The within sum of squares for cluster , SS
, is defined as follows:
![]() |
Note that to check the stability of the clustering it is desirable to
rerun the method, such that a new initial configuration is used.
Since the number of clusters is also unknown, it is also a good
idea to rerun the method using different values of .
According to [1] there are strong arguments for not
fixing the number of clusters in advance:
Bjørn Kåre Alsberg 2006-04-06