基于SOM聚类的可视化方法及应用研究

摘要

This paper proposed an unsupervised SOM-based cluster method for dealing with financial data. The clustered resultwas visualized on two dimensions by using parallel coordinate and interactive circular parallel coordinate. With this method,itcan form a clear visualization of the clustered result, improve the effect of visual clustering and enable user to reveal underlyingpatterns.%提出了用无监督的自组织映射方法对金融数据进行聚类,并用平行坐标和交互式的圆形平行坐标方法在二维平面上表示出来.用这种方法形成清晰的可视化聚类结果,不仅有效地总结了数据特征,还提高了聚类的可视效果,从而便于发现数据的变化趋势.