Machine learning (or statistical learning) describes the advanced use of algorithmic and statistical techniques to classify, predict, or structure large and often unstructured datasets. In recent years, its use (and usefulness) has exploded across industry, academia, government, and non-profit. Propelled by advances in computing power, better data availability, and more accessible programming tools, social science scholars are increasingly turning to ML to unlock new knowledge in fields such as political science, international development, economics, and psychology. 

This course will equip students with foundational knowledge of the main ideas, models, tools, and applications of ML for the social sciences. After providing the theoretical and conceptual underpinnings necessary to understand and critically review various commonly used ML techniques, students will present and discuss a selection of seminal and groundbreaking applications. In parallel, they will also be taught to implement in RStudio various predictive algorithms and ML tools such as regularized regressions, classification trees and clustering techniques through basic examples. 

Throughout the course, particular attention will be paid to comparing advantages and disadvantages of different approaches and discussing the potential pitfalls and ethical risks of applying ML in social science research.

Assessment

• Class participation: 20%

• Policy brief evaluation (group work): 20%

• In-class presentation (group work): 20%

• Final research proposal: 40%