-- KathrynStoltzfus - 31 Jan 2018

Selecting variables potentially associated with a variable of interest is an important step toward successes of inferences and correct identification of risk factors. Variable selection itself is of great interest and it is also an essential component in the construction of Bayesian networks (directed acyclic graphs) and testing of differential Bayesian networks. In some situations, the association can be reasonably described by parametric models such as linear regressions. In other situations, the association is non-linear in an unknown form and semi-parametric models are often applied to model the association. In this talk, I present methods developed recently that select variables in linear models and semi-parametric models, and methods for comparing Bayesian networks. The methods are illustrated using gene expression data and epigenetic data.