The Daily Bulletin Archive

Please join us for the applied mathematics talk "Penalization Techniques for Latent Class Regression," presented by Anthony Franklin, a Ph.D. candidate at North Carolina State University, at 3:00 p.m. Friday, November 2, in Cleveland Hall 418.

Abstract: When data are believed to have hidden clusters, it is common practice to identify the clusters and then conduct analysis in each cluster separately. Methods have been proposed to perform the cluster and regression analysis simultaneously for a pre-specified number of clusters. This process has been referred to in literature as "latent variable regression analysis." This research investigates imposing penalization techniques to identify and collapse regression parameters that may be "shared" or redundant across clusters for a given model.