Data visualization for domain exploration: highly multivariate interaction techniques

  • Authors:

  • Martin Theus

  • Affiliations:

  • Data Warehouse Analyst, VIAG Interkom, Munich, Germany

  • Venue:
  • Handbook of data mining and knowledge discovery
  • Year:
  • 2002

Quantified Score

Hi-index 0.00

Visualization

Abstract

Linking and highlighting interactive statistical graphics increases the dimensionality of data that can be explored. For highly multivariate data (i.e., more than ten to twenty variables) insight into the data by linking low-dimensional plots can be limited. Thus the need for high-dimensional plots arises. These plots--for example, rotating plots (grand tour, projection pursuit), parallel coordinate plots, or mosaic plots--can incorporate up to ten or more variables in a single plot. Linked highlighting and alterations inside these plots (e.g., zooming, reordering, or sorting) offer high-dimensional insights into data sets. Multiple selections via selection sequences offer a convenient way of interacting with high-dimensional subsets of the data using low-dimensional plots.