DATA SCIENTISTS’ EPISTEMIC THINKING FOR CREATING AND INTERPRETING VISUALIZATIONS

Abstract

The purpose of the study was to understand the experiences of data scientists regarding common skills and strategies for interpreting and creating data visualizations. In this study, the participants were researchers in data science. The Delphi method was used to gather common processes of data visualization through three rounds of surveys called Delphi panels where responses from the previous panel were used to frame the questions on the next panel. Skills and strategies were identified after Delphi Panel 1 and then brought back to the participants in Delphi Panel 2 to rate the level of importance they attributed to those skills/strategies. Consensus was determined using a cut-off for the interquartile range for each skill/strategy, and overall group ratings were presented to researchers in Delphi Panel 3 for them to adjust their ratings as desired. This study provided empirical evidence for a consensus set of skills/strategies that data scientists engage in when interpreting and creating visualizations.