Support a Project
The Responsible Data Science Framework
Introduction
The Quick Reference Guide for the Framework for Responsible Data Science (RDS) offers concise, actionable steps to ensure ethical practices throughout the project lifecycle. It is a practical tool for aligning projects with the principles of Fairness, Transparency, Privacy, and Veracity.
Download the quick reference guide
RDS Quick Reference Chapter by Chapter
Identification
Objective: Develop a clear understanding of the problem and corresponding questions.
Key Actions: Define the project's problem, goals, scope, constraints, success criteria, and potential issues.
Download this pageData Collection & Processing
Objective: Guarantee quality control of data during gathering and processing.
Key Actions: Address potential biases, handle missing values, and document datasets meticulously.
Download this pageStoring & Transferring
Objective: Securely store and transfer data.
Key Actions: Choose appropriate storage solutions and ensure secure data transfer.
Download this pageAnalysis, Modeling, & Machine Learning
Objective: Analyze data and develop models to draw conclusions and support decision-making.
Key Actions: Choose and train models responsibly while considering biases and explainability.
Download this pagePresentation & Visualization
Objective: Present data and insights accurately and responsibly to support informed decision-making.
Key Actions: Create visualizations that convey information clearly.
Download this pageEnd-User Consumption
Objective: Ensure end-users can effectively and safely interact with the data science product.
Key Actions: Create an end-product that provides value to users. Responsibly handle user-generated data.
Download this pageMonitoring & Evaluation
Objective: Continuously monitor and evaluate deployed models.
Key Actions: Ensure models consistently meet goals and do not cause harm.
Download this pageThe Responsible Data Science (RDS) Framework is a dynamic, living document created by and for the data science community. It’s a practical tool that guides organizations and data professionals in applying the four RDS principles—Fairness, Transparency, Privacy, and Veracity—to their data science projects. Our framework maps out key considerations and steps to find the most ethical outcomes possible.
We use the framework in our own projects and have made it readily available to others. We envision a world where RDS is the standard across all industries. While the second version was published in early 2024, thanks to the work of our RD Working Group members and volunteers, we always strive to keep the framework relevant to the ever-evolving tech landscape.
DOWNLOAD THE FRAMEWORK
Disclaimer about contributors: The RDS Framework is a product of collaboration by luminaries and professionals from various industries. Inclusion as a “Contributor” recognizes one’s participation in the document but does not imply that individual’s endorsement of all sections of the Framework.