"Everything should be made as simple as possible, but not simpler." - Albert Einstein

Seminar Announcement

Penalized Maximum Likelihood Estimation and Variable Selection in Spatial Linear Regression

Tingjin Chu, M.S. Candidate, Department of Statistics, Colorado State University

Wednesday, March 31, 2010

3:00 p.m., room 006, Statistics Bldg

ABSTRACT

In the paper, we consider variable selection and parameter estimation for spatial linear model with a Gaussian error process. Both penalized maximum likelihood procedure and one-step sparse estimation are considered under the SCAD penalty function. Theoretical properties, in term of consistency, sparsity and asymptotical normality, are established for both methods under increasing domain. A computationally feasible algorithm is proposed for the one-step sparse estimation, followed by a simulation study.

Advisory Committee:

Dr. Haonan Wang, Advisor
Dr. Jun Zhu, Co-Advisor
Dr. Mary Meyer, Committee Member
Dr. Jie Luo, Electrical & Computer Engineering, Outside Member