針對雙樣本右設限資料，加權對數秩(weighted log-rank)檢定與加權Kaplan-Meier (weighted Kaplan-Meier)檢定是最常被使用來檢定兩個存活機率的分配是否相等的問題。因為這兩個檢定在針對不同的情況下各有優點，故很難在未知的情況下，先行挑選檢定，使其能夠擁有較大的檢定力。因此，為了要將此兩種檢定的優點結合，本論文中將著重於同時使用此兩種檢定的機動檢定。我們延續Chi與Tsai(2001)的想法，針對加權對數秩和加權Kaplan-Meier此兩種檢定的線性組合，使用交叉驗證挑選此線性組合的權重，並與Chi和Tsai提出的建立在相同權重的線性組合來做比較。透過模擬研究說明我們所提出方法的優越性，並且將此檢定方法應用於實際資料。 For the two-sample censored data problem, the weighted log-rank (WLR) tests and weighted Kaplan-Meier (WKM) test are commonly used for testing the equality of two survival distributions. Since each test has different advantages against various alternatives, it’s hard to decide in advance which of the tests can be used to gain more power when the alternative is unknown. Hence, in order to combine the advantages of these two classes of tests, a versatile test based on WLR test and WKM test is then proposed. We develop a cross-validation versatile test to select appropriate weights in combining WLR and WKM which differs from Chi and Tsai who suggested the equal weights. Some numerical experiments are performed for illustrating the superiority of the proposed method and then the proposed testing procedure is applied to two real data sets.
