Hierarchical clustering is a clustering technique that generates clusters at multiple hierarchical levels, thereby generating a tree of clusters.

Examples

Common methods include DIANA (DIvisive ANAlysis) which performs top down clustering (usually starts from the entire data set and then divides it till eventually a point is reached where each data point resides in a single cluster, or reaches a user-defined condition).

Another widely known method is AGNES (AGlomerative NESting) which basically performs the opposite of DIANA.

Distance metric& some advantages

There are multitude of ways to compute the distance metric upon which the clustering techniques divide/accumulate in to new clusters (as complete and single link distances which basically compute maximum and minimum respectively).

Hierarchical clustering provides advantages to analysts with its visualization potential, given its output of the hierarchical classification of a dataset. Such trees (hierarchies) could be utilized in a myriad of ways.

Other non-hierarchical clustering techniques

Other clustering methodologies include, but are not limited to, partitioning techniques (as k means and PAM) and density based techniques (as DBSCAN) known for its advantageous discovery of unusual cluster shapes (as non-circular shapes).

Suggested learning sources to look into

• Han, Kamber and Pei's Data Mining book; whose lecture slides and companion material could be found here.
• Wikipedia has an entry on the topic here.