加权壳近邻填充数学模型

基金项目:

国家自然科学基金

详细信息

Weighted Shell-Neighbor Imputation Methods

摘要: 提出加权壳近邻填充缺失数据数学模型，充分利用壳近邻填充选取近邻数据的特性，侧重于被重复选择到的近邻点，有效提高了填充效果. 还提出一种称为goodness的新评价方法，克服了均方根误差（RMSE）的弱点. 实验结果表明，提出的加权壳近邻填充数学模型比一般的近邻填充的效果好，而且goodness评价方法比RMSE更能分辨出填充算法的性能.
- 加权壳近邻填充
Abstract: Weighted methods are designed for improving Shell Neighbor Imputation, Called WSNI. They utilize the features of nearest neighbor selection of the WSNI. All multiple-selected nearest neighbors are stressed in the WSNI algorithms. This lads to pretty good performance.Experimental results show that the porposed algorithms have better performances than conventional neighbor imputation method.
- weighting shell-neighbor imputation