用逆高斯模型時競爭風險的估計

Abstract

本論文是考慮當競爭風險是獨立的逆高斯模型時所做的一些推論。我們使用EM演算法去做最大概似估計，以及以貝氏的觀點，當給定客觀的先驗分配時，用MCMC的方法去做估計。最後，用真實的資料和模擬的資料用我們的方法去做推論。

In this thesis, we consider inference for the poly-inverse Gaussian model, which arises in competing risk scenarios when the risks have independent inverse Gaussian distributions. This thesis proposes the maximum likelihood estimation of the parameters of the poly-inverse Gaussian distribution by EM algorithm and Bayesian inference with some non-informative prior density functions using the MCMC method. Real and simulation data sets illustrate our approach to inference.