Missing data imputation by utilizing information within incomplete instances
Missing data imputation by utilizing information within incomplete instances
复制标题
利用不完整实例中的信息进行缺失数据插补
DOI：
10.1016/j.jss.2010.11.887
复制
发表时间：
2011-03-01
影响因子：
3.5
通讯作者：
Zhu, Xiaofeng
中科院分区：
文献类型：
--
作者：
Zhang, Shichao;Jin, Zhi;Zhu, Xiaofeng
本文建议在估计缺失值时在不完整的实例（具有丢失值的实例）中使用信息。因此，一种简单有效的非参数迭代归因算法（称为NIIA方法）是为迭代推出缺失的目标值而设计的。 NIIA方法将每个丢失值划分几次，直到算法收敛为止。在第一次迭代中，所有完整的实例都用于估计缺失值。自第二次摘要迭代以来，使用了不完整实例中的信息。我们进行了一些用于评估效率的实验，并证明：（1）不完整实例中信息的利用是有好处的，可以轻松捕获数据集的分布； （2）NIIA方法在准确性上的表现优于现有方法，并且当数据集具有较高的比率时，显然会突出显示此优势。 （c）2010 Elsevier Inc.保留所有权利。
This paper proposes to utilize information within incomplete instances (instances with missing values) when estimating missing values. Accordingly, a simple and efficient nonparametric iterative imputation algorithm, called the NIIA method, is designed for iteratively imputing missing target values. The NIIA method imputes each missing value several times until the algorithm converges. In the first iteration, all the complete instances are used to estimate missing values. The information within incomplete instances is utilized since the second imputation iteration. We conduct some experiments for evaluating the efficiency, and demonstrate: (1) the utilization of information within incomplete instances is of benefit to easily capture the distribution of a dataset; and (2) the NIIA method outperforms the existing methods in accuracy, and this advantage is clearly highlighted when datasets have a high missing ratio. (C) 2010 Elsevier Inc. All rights reserved.