Comparison and analysis of clustering techniques

Clustering is in the eye of beholder. Researchers have proposed many principles and models whose corresponding optimization problem can only be approximately solved by even large number of algorithms. Each algorithm emphasizes how it is different from previous algorithms. In our experiment we are comparing hierarchical and partitional clustering in terms of efficiency, intra-cluster similarity, and stability, a desirable property of clustering is stability that is small change to data should not lead dramatically different clustering.