依賴設限下轉換模式的統計推論

Abstract

在這篇論文中，我們假設連續性事件中最後一個階段的時間服從轉換模式以及死亡服從依賴設限。我們將檢查包含Cheng Wei & Ying (1995) 在內等已經提出的方法在新的資料結構下的正當性，同時我們也提出一個新的估計參數的方法。這些方法的表現可以藉由模擬所觀察到，而對於新方法的理論證明也將會有簡短的介紹。

In the thesis, we assume that the last stage of duration time for successive events follow a transformation model. It is known that the death are subject to dependent censoring. We examine the validity of existing methods, including the method proposed by Cheng, Wei and Ying (1995), under the current data structure. We also propose a new method for estimating the regression parameter. Finite-sample performance of these methods are investigated via simulations. Theoretical justification of the method is also briefly discussed.