基于SOM和引力场聚类的金融数据可视化

摘要

平行坐标技术是信息可视化中重要的分析手段,可以实现多维数据在二维空间上的可视化.为了给用户提供一种快捷、方便的金融数据可视化及分析工具,提出一种基于引力场聚类的金融数据可视化方法.首先利用自组织映射(SOM)对初始金融数据进行分类,使每类数据都含有特定的经济意义；然后进行视觉聚类,利用引力场原理对每个类中的折线进行聚拢,对类与类之间进行排斥,再通过设置不透明度以及交互操作等手段对可视化结果进行增强.实验结果表明,该方法可以形成清晰的可视化聚类结果,便于发现数据的变化规律.%Parallel coordinates technique is an important analysis tool in information visualization. It provides an intuitive way to visualize the multidimensional data on two-dimensional space. In this paper, an approach of financial data visualization based on the gravitational field clustering is proposed. Firstly, self-organizing map (SOM) is used to classify the raw financial data; therefore each class of data contains specific economic significance. Then the gravitational field theory is used to congregate fold line in each class and meanwhile to set the exclusion between the classes. Finally, the visualization results are enhanced by setting the opacity and interaction. The experimental result shows that the proposed method forms a clear visual clustering result and discovers the variation law of data.