Linear Models are widely used for multivariate analyses. Here you will find some notes and fundamental concepts.

EM-algorithm is a method to find maximum likelihood estimates that can be applied in a variety of cases. It is particularly useful for model based clustering estimation and models with latent variables.