Interaction Techniques for Stacked-Dimension Visualizations
Stacked-dimension tables (SDTs) represent multidimensional data with nested tables, where each table shows two of the dataset’s dimensions. SDTs provide a comprehensive overview of all of the data and all of its dimensions – but overviews are just the starting point for exploration, and there is little information available to designers about how to support further interactions with SDTs. We worked with a crop-breeding research group to develop a stacked-dimension system that suited their complex multidimensional datasets, and to identify requirements for their analyses of differential gene expression across multiple genomes. Based on the requirements, we developed a new SDT system and several new interaction techniques that support the researchers’ needs to filter the data, reconfigure the visualization, provide data context, revisit previous configurations, and integrate findings into a broader workflow. To test the generalizability of our designs, we then extended the requirements and techniques in a second SDT system for a new domain (outcomes from a retirement-planning model) and carried out a small usability study with this system. Our evaluations show that the techniques are easily learned and understood by both domain experts and everyday users, and that they provide support for real-world exploration in stacked-dimension visualizations.
Download the document
Download the bibtex