A Nested Hierarchy of Localized Scatterplots
Martin Eisemann, Geórgia Albuquerque, Marcus A. Magnor · 2014
The simplicity and visual clarity of scatterplots makes them one of the most widely-used visualization techniques for multivariate data. In complex data sets the important information can be hidden in subsets of the data, often obscured in the typical projections of the whole dataset. This paper presents a new interactive method to explore spatially distinct subsets of a dataset within a given projection. Precisely, we introduce a hierarchy of localized scatterplots as a novel visualization technique that allows to create scatterplots within scatterplots. The resulting visualization bears additional information that would otherwise be hidden within the data. To aid the useful interactive creation of such a hierarchy of localized scatterplots by a user we display transitions between scatterplots as animated rotations in 3D. We show the applicability of our visualization and exploration technique or different tasks, including cluster detection, classification, and comparative analyses. Additionally, we introduce a new exploration tool which we call the cross-dimensional semantic lens. Our hierarchy of localized scatterplots preserves the visual clarity and simplicity of scatterplots while providing additional and easily interpretable information about local subsets of the data.