摘要
针对当平行坐标中存在大量交叉、重叠的线段时,很难直观理解数据的分布规律和特征的问题。提出了一种基于类区间的多维数据可视化方法。首先根据数据的分布特征,对数据进行聚类,然后根据每个类的权值大小确定各类区间的宽度,并根据类区间互斥原则重新调整类中心点的位置,使各个类区间相互独立。将该方法应用到农药残留检测标准数据的分析中,形成了较好的可视化效果,为该领域人员提供一种高效、简单的数据可视化和分析工具。
When there is substantial overlap in parallel coordinates, overlapping segment, it is difficult to intuitively understand data distribution regularity and characteristics of the problem. A kind of multidimensional data visualization method was put forward based on the class interval. First according to the distribution features of the data for the data clustering, then according to the size of the weights of each class to determine all kinds of the width of the interval, and according to the class interval mutex principle to adjust the position of the center of class, each class interval was made independent of each other. This method was applied to pesticide residue detection standard data analysis, to form a good visual effect, for the field staff to provide a highly efficient, simple data visualization and analysis tools.
出处
《系统仿真学报》
CAS
CSCD
北大核心
2013年第10期2418-2423,共6页
Journal of System Simulation
基金
"十二五"国家科技支撑计划项目(2012BAD29B01-2)
北京市属高等学校科学技术与研究生教育创新工程建设项目(PXM2012_014213_000037
PXM2012_014213_000079)
关键词
类区间
多维数据
聚类
平行坐标
农药残留数据
cluster region
multidimensional data
clustering
parallel coordinates
pesticide residue data
