Further Applications of a Particle Visualization Framework

Our previous work introduced a 3D particle visualization framework that viewed each data point as being a particle affected by gravitational forces. We showed the use of this tool for visualizing cluster results and anomaly detection. This paper generalizes the particle visualization framework and demonstrates further applications to both clustering and classification. We illustrate its usage for three new applications. For clustering, we determine the appropriate number of clusters and visually compare different clustering algorithms’ stability and representational bias. For classification we illustrate how to visualize the high dimensional instance space in 3D to determine the relative accuracy for each class. We have made our visualization software that produces standard VRML (Virtual Reality Markup Language) freely available to allow its use for these and other applications.

• Type: Conference
• Year: 2004
• Where: PAKDD
• Authors: Ke Yin, Ian Davidson
• Keywords: Data Mining, Dimensional Instance Space, PAKDD 2004, Particle Visualization Framework, Virtual Reality Markup