Partitional Clustering

Simple Solution:
Select a criterion and evaluate it for all possible partitions containing K clusters.
Problem:
Combinatorial blow-up of the number of partitions to be evaluated.

Optimized Approach:
Start with a initial partition and move objects so that the value of the criterion function improves.
Problem:
Might converge to local minima.

• Square error clustering methods
• Clustering by mixture decomposition
• Clustering by density Estimation and mode seeking
• Clustering by graph theory
• Nearest Neighbor clustering
• Fuzzy Clustering