The general process of partition-based clustering is iterative. The first step defines or chooses a predefined number of representatives of the cluster and updates the representative after each iteration if the measure for the clustering quality has improved. The following diagram shows the typical process, that is, the partition of the given dataset into disjoint clusters:
The characteristics of partition-based clustering methods are as follows:
The resulting clusters are exclusive in most of the circumstances
The shape of the clusters are spherical, because of most of the measures adopted are distance-based measures
The representative of each cluster is usually the mean or medoid of the corresponding group (cluster) of points
A partition represents a cluster
These clusters are applicable for small-to-medium datasets
The algorithm will converge under certain convergence object functions, and the result clusters are often local optimum
The k-means clustering...