VISUALIZING UNCERTAINTY WITH CHROMATIC ABERRATION

Abstract

In recent years an increasing array of research are being conducted by researchers in the field of uncertainty visualization that attempt to determine the impact of representations on users’ perception and evaluate its effectiveness in decision making. Uncertainties are often an integral part of data and by nature model predictions also contain significant amounts of uncertain information. In this study, we explore a novel idea for a visualization to present predictive model uncertainties using Chromatic Aberration (CA). We first utilized existing machine learning models to obtain predictive results and then visualized the data itself and its associated uncertainties with an artificially spatially separated channels of red, green, and blue color components. This chromatic aberration representation has been evaluated in a comparative user study. From quantitative analysis it is observed that user is able to identify targets in CA method more accurately than quickly than Value-Suppressing Uncertainty Palettes (VSUP) approach.