How Data Scientists Navigate the Outer Loop of Client Collaborations in Industry and Academia

• Activity Number: 4 - Theoretical and Empirical Contributions to Data Analytic Practice
• Type: Invited
• Date/Time: Sunday, August 8, 2021 : 1:30 PM to 3:20 PM
• Sponsor: ENAR
• Author(s): Sean Kross*
• Companies: The University of California San Diego
• Keywords: Data Science; Design; Workflow; Clients; Collaboration; Practice

Abstract

Data scientists often serve as collaborators and consultations on scientific projects. What does the workflow of a data scientist’s partnership with clients look like throughout the lifetime of a collaboration? To answer this question, we investigated the social and emotional workflow that consulting data scientists lead their clients through in the course of a data analysis. We interviewed ten data scientists in diverse roles across industry and academia about their experiences as consultants. We discovered that they work with clients in a six-stage outer-loop workflow: 1) building trust before a project begins, 2) orienting to client’s constraints, 3) collaboratively framing the problems, 4) bridging data science and domain expertise, 5) technical data analysis work, 6) helping clients cope with analysis results. This outer-loop expands what collaboration means in data science beyond the programming for data science workflow of acquiring, cleaning, analyzing, modeling, and visualizing data. We will discuss the implications of this work for data science education, parallels to design work, and unmet needs for supporting data scientists.