Eliciting Strategies and Tasks in Uncertainty-Aware Data Analytics

Nadia Boukhelifa (1, 2) , Marc-Emmanuel Perrin (2) , Samuel Hurron (2) , James Eagan (2)

1 INRA - Institut National de la Recherche Agronomique
2 Télécom ParisTech

Uncertainty plays an important and complex role in data analysis and affects many domains. To understand how domain experts analyse data under uncertainty and the tasks they engage in, we conducted a qualitative user study with 12 participants from a variety of domains. We collected data from audio and video recordings of think-aloud demo sessions and semi-structured interviews. We found that analysts sometimes ignore known uncertainties in their data, but only when these are not relevant to their tasks. More often however, they deploy various coping strategies, aiming to understand , minimise or exploit the uncertainty. Within these coping strategies, we identified five high level tasks that appear to be common amongst all of our participants. We believe our findings and further analysis of this data will yield concrete design guidelines for uncertainty-aware visual analytics.

Mots clés: information visualization; qualitative study; uncertainty; data science

Domaines: Interface homme-machine [cs.HC]

Nadia Boukhelifa, Marc-Emmanuel Perrin, Samuel Hurron, James Eagan. Eliciting Strategies and Tasks in Uncertainty-Aware Data Analytics. IEEE Conference on Visual Analytics Science and Technology (IEEE VAST 2016) [Poster Paper], Oct 2016, Baltimore (Maryland), United States. ⟨hal-01404022⟩