Clustering
A branch and bound cluster analysis tree is shown below. 
In this example, the items to be clustered have been given
4 digit descriptors or names, the first digit being the most
significant descriptor and the last digit the least significant,
thus the two items "1111" and "1112" both
fall into the category or cluster "111x", similarly,
the two items "1121" and "1122" both fall
into the category or cluster "112x"; and both these
clusters fall within the larger cluster "11xx".
In creating such a cluster analysis tree, one should (conventionally)
continue clustering until all items and categories fall into
one "super" cluster - in this example "1xxx".
Various processes exist for clustering items, including statistical
software of varying degrees of complexity and sophistication.
However the purpose for which one is clustering items should
always be borne in mind as this will have a profound effect on
the final clusters that result. |
