Hierarchical method of Clustering algorithm

Firstly, the hierarchical clustering algorithm in clustering is introduced. Hierarchical method is divided into the hierarchical clustering of cohesion and the hierarchical clustering of division.

Condensation method: Also known as the bottom-up method, first each object as a separate cluster, and then according to the nature and rules to merge the similar classes successively, until all the objects are merged into a cluster, or meet a certain termination conditions. The classical hierarchical condensation algorithm is represented by the Agnes algorithm, and the improved hierarchical condensation algorithm is mainly represented by Birch,cure,rock,chameleon. (Detailed in the following sections)

The method of splitting: Also called the top-down method, just as in the condensation method, the first is to treat all objects as a cluster, and then at each step, the upper class is split into smaller classes, until each class contains only a single object, or it satisfies a termination condition. The split algorithm produces the same set of classes as the Coalescing method, except that the sequence of the generation is completely reversed. The classic hierarchical splitting algorithm is represented by the Diana algorithm.

Hierarchical method of Clustering algorithm