Model-Based Clustering

Assume that all the variables come from a set of some statistical distribution (multivariate normal distribution), and find those distributions (by parameterizing covariance matrix)

Automatic selection of K, by choosing the number of clusters that maximizes BIC.