摘要
本文介绍了数据挖掘中不完整数据的研究现状及ICA与ViSOM的特点,提出了基于ICA与ViSOM的不完整数据的处理模型IVIS-IDH,研究了数据之间存在相关关系且为非高斯分布时不完整数据的处理方法,对缺失数据值的估计方法及其估计的残差进行了详细的讨论和分析,并在ViSOM基础上取得了不完整数据集的可视化分析结果,从而克服了S.Wang提出的不完整数据处理方法的不足。
The paper introduces the state of incomplete data as well as ICA's and ViSOM's characteristics, studies the method of incomplete data sets under the circumstances of that data remain dependent and non-Gaussian, discusses the method estimation of missing data,and analyzes the carried-out the residual analysis. And then based on ICA and Vi- SOM, a model named IVIS-IDH, is proposed in this paper. The proposed model can achieve the visualization of incomplete data sets based on ViSOM, so that it overcomes the remedy for handing of incomplete data proposed by S. Wang.
出处
《计算机科学》
CSCD
北大核心
2007年第7期174-177,共4页
Computer Science
基金
国家自然科学基金资助项目(10371135)
关键词
不完整数据
ICA
ViSOM
相关关系
高斯分布
Incomplete data, ICA (independent component analysis), ViSOM (Visualization-Induced Self-Organizing Maps), Correlation, Gaussian distribution