Introduction to Finite Mixture Models and Latent Class Analysis
Dott.ssa Silvia Columbu
Postdoctoral researcher in Statistics
Department of Mathematics and Computer Science, University of Cagliari
This course is intended as both a theoretical and practical introduction to finite mixtures as a model based clustering tool. We will introduce the concept of finite mixture of distributions for categorical as well as continuous variables. We will study the problem of maximum likelihood via expectation-maximization (EM) algorithms. We will start from simple data structures and then move to the introduction of regressors and to more complex model structures. We will also explore the problem of model selection and choice of the number of clusters. We will see practical applications with the aid of devoted packages in the statistical software R. Students will learn how to implement an analysis and how to read and interpret outputs and graphical representations of results.
Duration: 20 hrs, in the period going from 18th to 26th of February 2021, for a total of 4 CFR