| 標題主旨 |
講題: Model Selection for Generalized Estimating Equations Accommodating Drop-out Missingness
演講者：Yi-Hau Chen 程毅豪 (Institute of Statistical Science, Academia Sinica 中研院統計所)
Abstract:
Generalized estimating equations (GEE) have been a popular framework for marginal regression analysis of longitudinal/clustered data. Akaike Information Criterion and Cp types of model selection have been developed for GEE, but these methods become problematic when applied to data with missing observations, which commonly arise in longitudinal/clustered studies. We propose a model selection criterion for GEE analysis where the data are subject to drop-put or monotone missingness, and the missing probabilities are known or can be estimated. Simulation results demonstrate the satisfactory performance of the proposed method, and the inadequacy of naive use of the existing model selection methods ignoring the missing data. The utility of the new method is further illustrated by real data analysis. This is a joint work with Dr. Chung-Wei Shen at Academia Sinica.
時間：2011年6月14日(星期二) 14：00 ∼ 15：00
地點：推廣大樓3樓9313室
<< 歡迎 老師,同學 踴躍參與!! >>
|
