國立政治大學統計學系主講人：陳立榜博士 (Ph.D. University of Waterloo, Canada)
學 術 演 講
學 術 演 講
題 目：Analysis of Graphical Models with Complex Noisy Data
時 間：民國109年10月19日 (星期一) 下午1：30
Graphical models are useful in characterizing the dependence structure of variables. They have been commonly used for the analysis of high-dimensional data and the incorporation of regression models. Many estimation procedures have been developed under various graphical models with a stringent assumption that the associated variables must be measured precisely. In applications, this assumption, however, is often unrealistic and mismeasurement in variables is usually presented in data. In this talk, we investigate the high-dimensional graphical model with error-prone variables. We first consider the exponential family graphical models with mismeasurement. Specifically, we examine the bias of the estimators induced by measurement error and then propose valid estimation procedures to account for measurement error effects. Next, we incorporate the network structure and error-prone predictors with modeling survival data. Different from conventional survival analysis, our approach entails several important features, including (1) many covariates are inactive in explaining the survival information, (2) active covariates are associated in a network structure, and (3) some covariates are error-contaminated. Moreover, our proposed models significantly enlarge the scope of the usual survival models and have great flexibility in characterizing survival data. To address mismeasurement effects and select active covariates simultaneously, we develop a simulation-based three-stage procedure. Theoretical results are established for the proposed methods and numerical studies are reported to assess the performance of our proposed methods.