Visualizing temporal cluster changes using Relative Density Self-Organizing Maps

Authors:
Denny;Graham J. Williams;Peter Christen
Affiliations:
The Australian National University, School of Computer Science, Canberra, Australia and University of Indonesia, Faculty of Computer Science, Depok, Indonesia;Australian Taxation Office, Canberra, Australia;The Australian National University, School of Computer Science, Canberra, Australia
Venue:
Knowledge and Information Systems - Special Issue:Best Papers from the 12th Pacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD2008);Guest Editors: Takashi Washio, Einoshin Suzuki and Kai Ming Ting
Year:
2010

Citing 0
Cited 4

Analysis of cluster migrations using self-organizing maps

PAKDD'11 Proceedings of the 15th international conference on New Frontiers in Applied Data Mining
Self-organizing time map: An abstraction of temporal multivariate patterns

Neurocomputing
Decomposing the global financial crisis: A Self-Organizing Time Map

Pattern Recognition Letters
Letters: Clustering of the Self-Organizing Time Map

Neurocomputing

Quantified Score

Hi-index	0.00

Visualization

Abstract

We introduce a Self-Organizing Map (SOM)-based visualization method that compares cluster structures in temporal datasets using Relative Density SOM (ReDSOM) visualization. ReDSOM visualizations combined with distance matrix-based visualizations and cluster color linking, is capable of visually identifying emerging clusters, disappearing clusters, split clusters, merged clusters, enlarging clusters, contracting clusters, the shifting of cluster centroids, and changes in cluster density. As an example, when a region in a SOM becomes significantly more dense compared to an earlier SOM, and is well separated from other regions, then the new region can be said to represent a new cluster. The capabilities of ReDSOM are demonstrated using synthetic datasets, as well as real-life datasets from the World Bank and the Australian Taxation Office. The results on the real-life datasets demonstrate that changes identified interactively can be related to actual changes. The identification of such cluster changes is important in many contexts, including the exploration of changes in population behavior in the context of compliance and fraud in taxation.