Prediction Error Estimation for Regularized Functional Mixed Models

• Activity Number: 231
• Type: Contributed
• Date/Time: Monday, August 4, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Abolfazl Safikhani*+ and Tapabrata Maiti and Ping-Shou Zhong
• Companies: Michigan State University and Michigan State University and Michigan State University
• Keywords: Functional data analysis ; mixed models ; mean square error ; prediction band ; B-spline basis functions

Abstract

Functional mixed models have become increasingly popular in scientific studies. While the current literature is confined to mean parameter estimation, we consider the estimation of mixed effects that combine both fixed and random components. The mixed effects are mainly used for prediction and measuring prediction error is a well known problem. Along with confidence band calculation for the mean curves, we derive an asymptotically valid formula for measuring the prediction errors. We consider penalized likelihood based estimation and thus some new theoretical results are established. The numerical study shows the methods are promising.