Statistical Methods for Analysis of Potentially Censored Skewed Data with an Application to Health Care Costs
主 题: Statistical Methods for Analysis of Potentially Censored Skewed Data with an Application to Health Care Costs
报告人: Professor Xiao-Hua Andrew Zhou (Peking University and University of Washington )
时 间: 2016-10-21 15:00-16:00
地 点: 理科一号楼 1114( 数学所活动）
Many random variables have a skewed distribution with heteroscedastic variance.For example, the health care of a subject shares such the features. Due to these features, it is difficult to analyze such the data, such as prediction of health care costs for patients. In this talk, I will introduce some new methods for the analysis of such data. These new mehods include: (1) a semi-parametric two-part heteroscedastic transformation model, (2) a quantile regression model, (3) a non-parametric heteroscedastic transformation regression model, (4) a semi-parametric two-part mixed-effects heteroscedastic transformation model, (5) a double robust estimator of average causal treatment effect for censored medical cost data, and (6) a semiparametric regression model for lifetime medical cost.