Improving the visual analysis of high-dimensional datasets using quality measures

Abstract

Modern visualization methods are needed to cope with very highdimensional data. Efficient visual analytical techniques are required to extract the information content in these data. The large number of possible projections for each method, which usually grow quadratically or even exponentially with the number of dimensions, urges the necessity to employ automatic reduction techniques, automatic sorting or selecting the projections, based on their informationbearing content. Different quality measures have been successfully applied for several specified user tasks and established visualization techniques, like Scatterplots, Scatterplot Matrices or Parallel Coordinates. Many other popular visualization techniques exist, but due to the structural differences, the measures are not directly applicable to them and new approaches are needed. In this paper we propose new quality measures for three popular visualization methods: Radviz, Pixel-Oriented Displays and Table Lenses. Our experiments show that these measures efficiently guide the visual analysis task.

DOI: 10.1109/VAST.2010.5652433

Extracted Key Phrases

12 Figures and Tables

051015201520162017
Citations per Year

Citation Velocity: 7

Averaging 7 citations per year over the last 3 years.

Learn more about how we calculate this metric in our FAQ.

Cite this paper

@inproceedings{Albuquerque2010ImprovingTV, title={Improving the visual analysis of high-dimensional datasets using quality measures}, author={Georgia Albuquerque and Martin Eisemann and Dirk J. Lehmann and Holger Theisel and Marcus A. Magnor}, booktitle={IEEE VAST}, year={2010} }