Multivariate regression with variable selection using Gaussian process priors

In this talk, it will be considered the problem of variable selection in regression models with multiple responses and covariates.

It will be presented models to explore two sets of high-dimensional multivariate data in this setting with the goal of identifying clusters of responses that share the same relation to the same subsets of covariates. It will be presented a stochastic partitioning approach to identify linear associations between subsets of variables in the two datasets. Then we will discuss an extension that employs Gaussian process priors to model the mean of the responses as a function of the covariates in each cluster. The method will be illustrated on simulated data.

Zoom Link: https://kocun.zoom.us/j/93188815896

Meeting ID: 931 8881 5896

Password: 978133