Futzing and Moseying: Interviews with Professional Data Analysts on Exploration Practices

Area of Science:

• Data Science
• Human-Computer Interaction
• Information Science

Background:

• Data exploration is a critical phase in data analysis.
• Understanding practitioners' real-world approaches is essential for effective tool design.

Purpose of the Study:

• To investigate how professional data analysts conduct exploratory data analysis.
• To identify common tools and techniques used in data exploration.
• To understand diverse perspectives on the goals and methods of data exploration.

Main Methods:

• Qualitative study involving interviews with thirty professional data analysts.
• Focus on descriptions of exploratory activities and tool usage.
• Analysis of data from industrial, academic, and regulatory environments.

Main Results:

• Identified distinctions between exploration as a precursor to analysis versus open-ended exploration.
• Revealed conflicting views on whether "finding something interesting" is a valid exploration goal.
• Noted pervasive use of visualization, but limited use of direct manipulation interfaces.
• Observed varied opinions on the role of intelligent tools in data exploration.

Conclusions:

• Current data exploration practices are diverse and context-dependent.
• Future tool development should consider the spectrum of exploratory goals and methods.
• Further research can refine the definition and application of "data exploration" based on practitioner insights.