《Table 6 Mean Squared Error (MSE) Assuming Three Copula Types》

  提示：宽带有限、当前游客访问压缩模式

本系列图表出处文件名：随高清版一同展现

《双变量生存数据之间的关联评估——一种基于Copula的方法（英文）》

1. 获取 高清版本忘记账户？点击这里登录

1. 下载图表忘记账户？点击这里登录

From our simulation study，our normal-based estimation procedure always tends to have downward biased estimates when the underlying copula structure is Normal，Clayton，Gumbel and Frank.When the percentage of censoring increases to 30%，all the four estimation procedures tend to get less accurate estimates comparing to the setting of 0%censoring when the underlying copula structure are Normal，Clayton，Frank and Gumebl.However，Gumbel-based estimation procedure is much more robust for data from Gumbel copula in terms of percentage of censoring.Finally，we also found that Frank，Gumebl，and Clayton-based estimation procedures perfom better with the increase of a positive association under the condition that we choose the right copula-based estimation procedure which means the copula assumption we made is the same as the true one.