Understanding the Role of Alternatives in Data Analysis Practices

Area of Science:

• Data Science
• Human-Computer Interaction

Background:

• Many professionals analyze data without identifying as data scientists.
• These data workers explore various hypotheses, data sources, algorithms, methods, tools, and visual designs, termed 'alternatives'.

Purpose of the Study:

• To understand why data workers explore alternatives.
• To characterize notions, processes, and strategies for exploring alternatives in data analysis.
• To develop a framework for understanding and supporting data workers' exploration of alternatives.

Main Methods:

• Conducted semi-structured interviews with 12 data workers.
• Performed four types of analyses on interview data.
• Developed a framework based on attention, abstraction, and analytic processes.

Main Results:

• Data workers' expertise, tool experience, and collaboration influence alternative exploration.
• Identified key aspects of why, how, and when data workers explore alternatives.
• Found a need for improved tools to support alternative exploration and management.

Conclusions:

• A framework is proposed to understand data workers' consideration of alternatives.
• The framework can guide the design of better data analysis tools.
• Further research is needed to deepen the understanding of alternatives in data work.