Final Exam - Jaewon Lim
Committee: Alex Luedtke (chair), Marco Carone, Ting Ye, Abraham Flaxman (GSR)
Presentation: Orthogonal statistical learning and inference for function-valued parameters with heterogeneous data sources
Abstract: There is a growing literature on estimating function-valued parameters by combining semiparametric efficiency theory and statistical learning. A second trend is the integration of multiple, partially overlapping data sources through data fusion and two-phase sampling. This dissertation develops theory and methodology at the intersection of these two directions.
In the first aim, I study data fusion for estimating causal dose-response functions, where a single dataset does not contain enough observations at every exposure level. I introduce a data fusion framework leveraging partially aligned sources, and construct a Neyman-orthogonal loss robust to nuisance estimation error. I develop two estimation algorithms, a general two-stage procedure and a closed-form kernel ridge regression estimator and…
Event interval: Single day event. Campus location: Hans Rosling Center for Population Health (HRC). Online Meeting Link: https://washington.zoom.us/j/92798528454. Campus room: HRC 375. Accessibility Contact: Deb Nelson, nelsod6@uw.edu, 206-685-9323. Event Types: Academics.
Tuesday, August 18, 2026, 10:30 AM – 12:30 PM.