New aspects of statistical methods for missing data problems, with applications in bioinformatics and genetics

摘要: We propose a nonparametric imputation procedure for data with missing val­ ues and establish empirical likelihood based inference parameters defined by general estimating equations. The is carried out multiple times via nonparamet­ ric estimator of the conditional distribution component given always observable random vector under study. used to construct profile parameter interest. demonstrate that proposed can correct selection bias in missingness leads more efficient estimation. method evaluated simulation an study on relationship between eye weight gene transcriptional abundance recombinant inbred mice.