In this project, the investigators study the fundamental aspects of incorporating uncertainty with sensitivity analysis in the visual analytics process. They also aim to develop novel and scalable visual representations of sensitivity, from the visualization of the raw sensitivity coefficients to visual summaries of multivariate derivatives obtained from the analysis. Uncertainty-aware visual analytics helps enhance analysts' confidence levels on the insight gained from the analysis. Furthermore, it gives toolmakers a methodology for measuring and comparing the robustness of data and visual transformations. Sensitivity coefficients of data and visual transformations are useful for discovering the factors that mostly contribute to output variability, identifying stability regions of the different transformations within the original data space, and telling the analyst what the interaction is between variables, outputs and transformations.
Uncertainty is introduced throughout the process of data generation, transformation, and analysis in most real-world applications. The ability to incorporate uncertainty into visual analysis is therefore critically important for insightful reasoning and key decision making. This project will have a wide-reaching impact on those areas relying on the ability to reason about large amounts of data. On one hand, it suggests to provide a variational view of the visual analytics process, which opens up new directions and paradigms for visual data analysis and mining. On the other hand, the improved understanding of the visual analytics process will help establish the field as a scientific discipline.
