Model Selection and Estimation for Semiparametric Stochastic Mixed Models for Longitudinal Data

• Activity Number: 464
• Type: Topic Contributed
• Date/Time: Wednesday, August 1, 2007 : 2:00 PM to 3:50 PM
• Sponsor: Section on Physical and Engineering Sciences
• Author(s): Xiao Ni*+ and Daowen Zhang and Hao Zhang
• Companies: North Carolina State University and North Carolina State University and North Carolina State University
• Keywords: correlated data ; Gaussian stochastic process ; variable selection ; smoothly clipped absolute deviation ; smoothing splines

Abstract

We propose a double-penalized likelihood approach for simultaneous model selection and estimation in semiparametric stochastic mixed models for longitudinal data. Two penalties are imposed on the ordinary likelihood: the roughness penalty for the nonparametric baseline function and the shrinkage penalty for the linear coefficients. We present an algorithm for maximizing the double-penalized likelihood and solving for various model components and unknown parameters within a modified linear mixed model framework. We propose and compare frequentist and Bayesian inference for model parameters. Simulation results are given to show effectiveness of our method. A real data example is also given to illustrate the use of our method.