多维数据的不确定性可视相关分析-计算机辅助设计与图形学学报.PDFVIP

30 6 Vol.30 No.6 2018 6 Journal of Computer-Aided Design Computer Graphics Jun. 2018 多维数据的不确定性可视相关分析 张 怡, 熊朝阳, 张加万 (天津大学软件学院 天津 300350) (yizhang@tju.edu.cn) : 基于不确定性的相关关系和原始数据的相关关系作为 2 种相关关系具有一定的联系与区别. 为了找出不确 定性相关关系并有效地对比这 2 种关系的异同, 发现可能产生额外不确定性的数据特征, 提出了一套分析框架和交 互系统. 首先进行聚类分析, 并根据聚类结果得到分组; 然后根据不确定性定量方法计算得到的每组不确定性数据 来构建不确定性数据集, 并进行相关分析和对比分析; 最后建立可视分析系统, 帮助用户利用不确定性相关关系来 筛选出可能产生额外不确定性的数据. 使用2 个差异较大的真实数据集, 验证了文中框架和系统的可用性和有效性. : 不确定性可视化; 多维数据; 数据聚类; 相关分析 : TP391.41 DOI: 10.3724/SP.J.1089.2018.16643 Uncertainty-Aware Visual Correlation Analysis for Multidimensional Data Zhang Yi, Xiong Zhaoyang, and Zhang Jiawan (School of Computer Software, Tianjin University, Tianjin 300350) Abstract: There are some similarities and differences between the correlation based on uncertainty and the correlation of original data. In order to find out the uncertainty correlation, effectively compare the similari- ties and differences between the two correlations and discover the data characteristics that may generate ad- ditional uncertainty, this paper proposed a framework for analyzing the uncertainty correlation and an inter- active system. Firstly, the framework performs clustering analysis, and obtains several groups according to the result of clustering analysis. Then it constructs the uncertainty data set through calculating the uncer- tainty of each group by the uncertainty quantification method and executes the correlation analysis and comparative analysis. Finally, a visual analysis system is established to help users use the uncertainty corre- lation to filter out data that may generate additional uncertainty. Experimental resu