中文摘要：

由于失访、改变治疗方案、研究时间结束时事件尚未发生等情况，所采集的数据中许多应采集而未能采集、应提交而未在一些时点上提交，造成数据不完全，这便产生了删失数据。因此研究删失数据的大样本性质是解决实际问题的关键。本文在｛Xi ，i≥1｝是α混合的随机变量，｛Yi ，i≥1｝是独立同分布的随机变量，且Xi与Yi相互独立的情形下，研究了随机删失数据下概率密度函数的核估计，获得了此核估计的逐点强相合速度和一致强相合速度。

英文摘要：

Censoring occurs for many reasons,such as loss of follow-up,changes in treatment and events not happened at the end of experiment.In such cases,the value of a measurement or observation is only partially known.Different from traditional large-sample analysis,special techniques should be used to handle censored data.In this paper,{Xi ,i≥1}is a set of mixed random variables,{Yi ,i≥1}is identically distributed ran-dom variables.{Xi ,i≥1}and{Yi ,i≥1}are independent.We established a kernel estimator of the density function for censored data,and get the result,namely pointwise convergence rate of strong consistency of kernel density estimation and uniform convergence rate of strong consistency of kernel density estimation.