Java: Data Science Made Easy

Hierarchical methods

Hierarchical methods are considered the simplest clustering algorithms; they are easy to understand and interpret. There are two families of clustering method, which belong to the hierarchical family:

• Divisive clustering algorithms
• Agglomerative clustering algorithms

In the divisive approach, we put all the data items into one cluster, and at each step we pick up a cluster and then split it into halves until every element is its own cluster. For this reason, this approach is sometimes called top-down clustering.

The agglomerative clustering approach is the opposite; at the beginning, each data point belongs to its own cluster, and then at each step, we select two closest clusters and merge them, until there is only ...