Model-Based Clustering

This chapter deals with the model-based approach to clustering based on a probability model, specifically the Finite Mixture Model (FMM). Within such a framework, the crucial problems of determining the number of clusters and choosing an appropriate clustering method can be seen as a problem of model selection. In this chapter, after a brief introduction to the model-based clustering framework, FMMs will be introduced with particular emphasis on the most commonly used family, namely, the Gaussian, for continuous data. Then, we briefly mention some approaches to model-based clustering for non-continuous data. The analysis of several real-world data will also be discussed.