The Need for an Enhanced Process Methodology for Ethical Data Science Projects

Abstract

This paper proposes a practice-based approach to ethics by expanding the scope of process framework CRISPDM for managing ethical risks arising in data science projects. The paper discusses the benefits and shortcomings of CRISPDM with respect to minimizing ethical risks during data science project execution. Next, to help ensure ethical data science projects, CRISP-DM is enhanced to minimize ethical risks. This refinement offers viable and specific work actions for data science practitioners. Furthermore, future research to integrate this enhanced CRISP-DM with a collaboration framework such as Scrum or Data Driven Scrum (DDS) is proposed to enable better coordination and efficient testing before final deployment of data science system.