Interaction with Uncertainty in Visualisations

Abstract

In recent years, a number of algorithms have been developed which provide fast approximate computations on large datasets. There is considerable interest in exploiting these techniques to render interactive visualisations of large datasets. Such interactions require careful consideration of the uncertainty arising from the approximations that these algorithms employ. Characterising and comparing visualisations of uncertainty has been well studied; however, it is typically assumed that uncertainty is a fixed property of the data. In light of this new generation of approximate visualisations, we consider the case where uncertainty arises from algorithms whose parameters can be altered, and so can be manipulated just like any other aspect of the visualisation. We present a novel directmanipulation interface for uncertainty in visualisations and show through a user study that our interface enables people to successfully edit and comprehend uncertainty.