Department of Statistics Seminar
North Carolina State University

presents

Chong Gu

Purdue University
(visiting NCSU and NISS)

"Penalized Likelihood Function Estimation: Probability Density and Hazard Rate"

ABSTRACT

The penalized likelihood method estimates a function of interest f by minimizing a score of the form L(f)+J(f), where L(f) measures the lack of fit such as the minus log likelihood, and J(f) is a roughness penalty penalizing the irregularity of the estimate. In this talk, we will illustrate the application of the method in some density and hazard estimation problems. For density estimation, we will show how to handle various kinds of sampling bias and how to estimate conditional densities. For hazard estimation, we will show how to use penalized full likelihood to estimate a proportional hazard model and beyond.

Friday, November 14, 1997

3:35 - 4:35 pm

206 Cox Hall

Refreshments will be served on the second floor of Dabney Hall (left of Room 222) at 3:00 pm.