Constructing overview + detail dendrogram-matrix views

Jin Chen, Alan M. MacEachren, Donna J. Peuquet

Research output: Contribution to journalArticle

15 Scopus citations

Abstract

A dendrogram that visualizes a clustering hierarchy is often integrated with a reorderable matrix for pattern identification. The method is widely used in many research fields including biology, geography, statistics, and data mining. However, most dendrograms do not scale up well, particularly with respect to problems of graphical and cognitive information overload. This research proposes a strategy that links an overview dendrogram and a detail-view dendrogram, each integrated with a re-orderable matrix. The overview displays only a user-controlled, limited number of nodes that represent the "skeleton" of a hierarchy. The detail view displays the sub-tree represented by a selected meta-node in the overview. The research presented here focuses on constructing a concise overview dendrogram and its coordination with a detail view. The proposed method has the following benefits: dramatic alleviation of information overload, enhanced scalability and data abstraction quality on the dendrogram, and the support of data exploration at arbitrary levels of detail. The contribution of the paper includes a new metric to measure the "importance" of nodes in a dendrogram; the method to construct the concise overview dendrogram from the dynamically-identified, important nodes; and measure for evaluating the data abstraction quality for dendrograms. We evaluate and compare the proposed method to some related existing methods, and demonstrating how the proposed method can help users find interesting patterns through a case study on county-level U.S. cervical cancer mortality and demographic data.

Original languageEnglish (US)
Article number5290691
Pages (from-to)889-896
Number of pages8
JournalIEEE Transactions on Visualization and Computer Graphics
Volume15
Issue number6
DOIs
StatePublished - Nov 1 2009

    Fingerprint

All Science Journal Classification (ASJC) codes

  • Software
  • Signal Processing
  • Computer Vision and Pattern Recognition
  • Computer Graphics and Computer-Aided Design

Cite this