Grounded Visual Analytics: A New Approach to Discovering Phenomena in Data at Scale

Abstract

We introduce Grounded Visual Analytics, a new method that integrates qualitative and quantitative approaches in order to help investigators discover patterns about human activity. Investigators who develop or study systems often use log data, which keeps track of interactions their participants perform. Discovering and characterizing patterns in this data is important because it can help guide interactive computing system design. This new approach integrates Visual Analytics, a field that investigates Information Visualization and interactive machine learning, and Grounded Theory, a rigorous qualitative research method for discovering nuanced understanding of qualitative data. This dissertation defines and motivates this new approach, reviews relevant existing tools, builds the Log Timelines system. We present and analyze six case studies that use Log Timelines, a probe that we created in order explore Grounded Visual Analytics. In a series of case studies, we collaborate with a participant-investigator on their own project and data. Their use of Grounded Visual Analytics generates ideas about how future research can bridge the gap between qualitative and quantitative methods.