RSEConUK 2019

There are many different representations of data values that are used in visualisations, but it is less common to see visual representations of uncertainty, especially in visualisations intended for non-technical audiences.

We have attempted to develop a visualisation method that relates uncertainty to visual complexity. We have constructed a scale of glyphs designed to work in 2D and 3D environments using R, Python and Blender, an open source 3D creation suite.

One motivation for this work is the representation of data from a network of IoT sensors. The sensors report a continuous stream of values, which is often summarised as the average over an hour. However, the quality of the sensors is not consistent and so it was necessary to find a simple but consistent way to represent the variance of the readings, so that users can understand the level of uncertainty associated with each sensor’s readings at a glance.

We will discuss the results from a series of trials designed to evaluate how users respond to the glyphs, including their ability to determine whether one glyph is more or less uncertain than another, as well as whether they are able to identify a specific glyph in context (i.e. when displayed on a map).

We submitted this talk because the visualisation of uncertainty is increasingly relevant in a number of fields and we feel that this talk will be suitable to RSEs from a range of background and subject areas, particularly those working on (or interested in) data visualisation.