Median regression models with censored data as often used models haven been discussed in recent years, when the censoring distribution and the covariates are independent, it is often used the Kaplan-Meier estimator to estimate the censoring distribution. When the censoring distribution depends on the covariates, there is few articles to discuss the problem. Specifically, as there is error margin of the measured covariates, there is nothing. In the second chapter, I propose semi-parametric estimate to get median regression models with the measured error margin and use it to estimate the mediant of survival data. In the third chapter, I propose the generalized product-limit estimator for the estimating survival function, which based on weighted estimating equation. And we discuss about the congruence of the censored data survival function using experienced survival function and sub-experienced function, furthermore, it's convergent. We conclude that the censored data survival function have good large sample qualities under some conditions.
|