Column-based Cluster and Bar Axis Density in Parallel Coordinates
In this paper we organize multi-dimensional datasets with column-based approach instead of the traditional row-based method, each column referring to one dimension and we use bar axis in place of line axis to represent corresponding dimension. Then parallel coordinates with column-based cluster, bar axis density and other techniques is used to convey a large complex multi-dimensional dataset in a relative small screen through the following steps: (a) visualization of column-based clusters with user-defined granularity to simplify the corresponding dimension where we group all the data points into several discrete values; (b) several distinct colors to distinguish the lines contain different amount of data points; (c) opacity is introduced to visualization to tell the difference among the lines with the same color; (d) brand instead of polyline to reveal the centre and the extent of each cluster; (e) layer-based drawing technique to emphasize the heavy lines and to denote the trend of multi-dimensional datasets; (f) bar axis to provide special space to illustrate the density of the dataset on each axis. Anyway, our work has two primary goals: one is to convey large dataset with legible compact vivid visualization on a limited screen area. The other one is to simultaneously reveal as many information features as possible away from clutter.
information visualization parallel coordinates cluster k-means density brand
Lei Tang Xue-qing Li Wen-jing Qi Zhi-fang Jiang
School of Computer Science & Technology,Shandong University,Jinan, 250101, China
国际会议
The 3rd Visual Information Communication-International Symposium(第三届视觉信息通信国际研讨会VINCI 2010)
北京
英文
27-34
2010-09-28（万方平台首次上网日期，不代表论文的发表时间）