Malhotra (2008) divides the cluster analysis methods into two subcategories: hierarchic and non-hierarchic.
Of the two methods, the non-hierarchic clusters are more suitable for dealing with big samples.
The non-hierarchic methods, K-middle methods, can be analyzed by measuring the threshold value.
With this method the number of the formed clusters has to be defined in advance, and the middle of the cluster is supposed to be known at the outset (though these initial cluster centers are temporary).