Single Cluster Visualization to Optimize Air Traffic Management

In this paper we present an application of single cluster visualization (SCV) a technique to visualize single clusters of high-dimensional data. This method maps a single cluster to the plane trying to preserve the relative distances of feature vectors to the corresponding prototype vector. Thus, fuzzy clustering results representing relative distances in the form of a partition matrix as well as hard clustering partitions can be visualized with this technique. The resulting two-dimensional scatter plot illustrates the compactness of a certain cluster and the need of additional prototypes as well. In this work, we will demonstrate the visualization method on a practical application.

• Type: Conference
• Year: 2006
• Where: GFKL
• Authors: Frank Rehm, Frank Klawonn, Rudolf Kruse
• Keywords: Data Mining, GFKL 2006, Relative Distances, Single Cluster, Single Cluster Visualization