EVENTS

Seminar: Functional Linear Models with Longitudinal or Functional Covariate Processes (Naisyin Wang)

Seminar

Department of Statistics

Thursday, 15 October 2009 at 3:30PM

Sidney Smith Hall, Room 1086

Functional Linear Models with Longitudinal or Functional Covariate Processes

Professor Naisyin Wang

University of Michigan

It is common to have a regression models with multiple longitudinal processes as covariates and a primary scalar response. We propose estimation procedures and corresponding supportive theory that allow one to perform investigation without making unnecessary distributional assumption on unobserved latent features in the longitudinal or functional covariates. The uncertainty which is associated with accounting for these latent features is also properly taken into consideration. We investigate the practical implications behind certain theoretical assumptions, which aim at having a better understanding of where the estimation variation lies. Numerical performances of the proposed approach were illustrated through simulations and practical examples.