Introduction to Pattern Recognition: A Matlab Approach

7.7 Hierarchical Clustering Algorithms 199

The two main categories of hierarchical clustering algorithms are:

• Agglomerative. Here the initial clustering 

0

consists of N clusters, each containing a single element

of X. The clustering 

1

, produced at the next step, contains N −1clustersand

0

is nested in it.

Finally, the 

N−1

clustering is obtained, which contains a single cluster (the whole data set X).

• Divisive. Here the reverse path is followed. The initial cluster 

0

consists of a single cluster (the

whole data set X). In the next step the clustering 

1

is produced, which consists of two clusters and is

nested in 

0

. Finally, the 

N−1

clustering