K2SP 3015 - Data and algorithms for Public Policy

This course explores how data-driven and algorithmic systems are transforming the way

public policies are designed and implemented. From social-benefit allocation and education

planning to fraud detection, governments increasingly rely on predictive models, scoring

systems, and AI tools to support decisions. These technologies promise efficiency and

consistency, but they also raise new challenges for accountability, fairness, and democratic

oversight.

Adopting a public-governance and applied perspective, the course examines how algorithmic

systems are conceived, commissioned, and used within public administrations. Students

analyse how these systems interact with policy goals, institutions, and values, and how they

redistribute responsibilities among public officials, private actors, and citizens.

Students are expected to recall the basics of Python. If a refresher is needed, we recommend taking the course Data Analysis with Python - Advanced” as option one. The course “Data Analysis with Python - Beginner” may also be considered upon request.

This course combines conceptual discussions and hands-on work across different formats:

• Conceptual sessions (2 hours each)

These sessions focus on the presentation and discussion of key concepts, frameworks, and policy questions related to data and algorithms in the public sector. Each session is dedicated to a specific topic.

• Practical sessions (3 hours each)

These sessions are devoted to applied work on data and algorithms, based on concrete use cases or technical challenges. They are designed to develop students’ ability to engage with real-world systems and datasets.

The course will also feature guest speakers, including civil servants and data scientists working in French public administrations, as well as practitioners from European institutions and think tanks. Their contributions will provide additional insights into how algorithmic systems are designed, implemented, and governed in practice.

Group project (45%) : in-depth investigation of a real-world algorithmic system used in the public sector made of an intermediate short oral presentation and a final written report

Individual exam (45%) : 2 hours exam in class combining theoretical questions on key course concepts and practical exercises in Python focused on data analysis

Participation and completion of class activities 10%

For selected 2-hour session, students will prepare by reading a short paper or article. Some sessions will also include a short post-session question (submitted by email or on the forum) to consolidate learning and support reflection.