Copula模式之下 雙變元存活資料之統計推論

Abstract

本論文主要是考慮在Copula模式之下,雙變元存活資料受限於右設限之半母數推論.目前存在許多估計相關性參數之半母數推論方法.我們比較三種推論方法針對不存在解釋變數之同質性資料.並利用迴歸的概念將現有的方法延伸至處理邊際異質的資料.最後,藉由模擬結果檢視上述方法之有限樣本的表現.

The thesis considers semi-parametric inference based on Copula models for bivariate survival data subject to right censoring. There exist several semi-parametric inference approaches to estimating the association parameter. We examine and compare three approaches developed for homogeneous data in absence of covariates. Then we extend these methods to a regression setting that accounts for marginal heterogeneity explained by the covariates. Finite-sample performances are examined by simulations.