A Decluttering Lens for Scatterplots

Molchanov, Vladimir*; Rave, Hennes*; Linsen, Lars

Abstract

Scatterplots are among the most popular visualization methods for bivariate or multivariate data. While scatterplots scale well with the number of samples, visual clutter cannot be avoided with increasing data size. Density regularization is a common approach to declutter scatterplots. However, most existing density-equalizing algorithms are restricted to rectangular domains, which limits their applicability. In particular, they cannot operate within interactive lenses, a widely used approach for interactive data exploration. We present a numerical approach that generalizes density-equalizing transformations to domains of arbitrary shape, including concave regions. The definition of the regions of interest can be data-driven or interactive. We demonstrate the effectiveness of our method by implementing adaptive and flexible interactive lenses for enhanced data exploration in scatterplots, showcasing its versatility and potential for broader application.