二維存活資料之模式檢驗

Abstract

在本論文中，我們針對右設限的資料提出了Archimedean Copula(AC)模型的模式檢驗法。我們拓展了Shih(Biometrika,1998) 的想法，Shih只針對Clayton模型作推論，而我們將之延伸到更大的集合，AC家族。我們也針對AC家族提出了新的資料生成演算法。我們提供模擬分析以佐證在有限樣本下，我們所提出的方法之效能。

In the thesis, we propose a model diagnostic approach to selecting an Archimedean Copula (AC) model based on right censored data. The proposed method extends the idea of Shih (Biometrika, 1998), who considered the Clayton model, to a larger class of models, namely the AC family. We also propose a new algorithm for generating a model from the AC family. Simulation results are provided to examine finite-sample performances of the proposed method.