Statistics 797L - ST-Mixture Models

A challenging task of statistical analysis is to identify hidden patterns in heterogeneous data and to understand/explain part of them using statistical models. Latent variable mixture models provide a natural framework for studying heterogeneity by classifying individuals into unobserved or latent groupings/classes with more homogeneous patterns. This course will focus on introducing the ideas and theories of latent variable mixture models and their applications in various areas such as biology, medicine, genomics, epidemiology, toxicology, pharmacokinetics, psychophysiology, social and psychological sciences, marketing, economics and finance. Topics will include the (parametric/semiparametric/nonparametric) model specification, identifiability, estimation methods, large sample properties, finite sample properties (using convex geometry), computational algorithms, data analysis, and interpretation).

This course will be accessible to students with knowledge of statistics at an intermediate level (i.e., probability, basis for the statistical inference and regression analysis).

Open to Graduate students only. STATISTC 607 & either 625/697R STATISTC 608 is a co-requisite for this course.