半参数方法在缺失数据中的研究及应用

Research and Application of Semi-parametric Methods in Missing Data

为了更多的利用缺失数据提供的信息,采用经验似然给出缺失协变量的条件分布的半参数方法对数据的缺失部分进行处理,并与几种比较常见的数据缺失处理方法——均值插补、多重插补、EM方法等进行了比较。模拟和实证研究中,在不同的缺失率下,运用Logistic模型,对几种方法的缺失处理效果加以比较,结果表明,随着缺失率的增加,几种方法的性能都受到影响,而半参数方法较其他几种方法更好。

In order to make more use of the information provided by missing data,the semi-parametric method of conditional distribution of missing covariates is used to process the missing part of the data and compare with several common data missing processing methods--mean interpolation,multiple interpolation,EM. In the simulation and empirical studies,the Logistic model is used to compare the effects of the missing methods at different missing rates. The results show that the performance of several methods is affected with the increase of the missing rate,the semi-parameters method is better than the other methods.