This course explores how modern developments in machine learning, AI and statistical learning may be leveraged in order to improve regression analysis and reduced-form econometrics more generally. For instance, we provide a modern, optimization-conscious treatment of linear regression, quantile regression, instrumental variables, and Fréchet-Hoeffding bounds. Throughout, we make an effort to produce methods that remain valid even when models are --as they always are-- misspecified. To produce such regression methods, we fetch tools and results from fields such as linear programming, optimal transport, numerical linear algebra, deep learning and reinforcement learning.
Quarter | Title | Instructor | Day(s) | Time(s) | Syllabus |
---|---|---|---|---|---|
Fall 2024 | Modern Methods for Applied Regression | Guillaume Pouliot | Tuesday, Thursday | 3:30pm-4:50pm | Syllabus |
Fall 2024 | Modern Methods for Applied Regression | Guillaume Pouliot | Tuesday, Thursday | 5:00pm-6:20pm | Syllabus |