Statistics 797S - ST-Estimation/SemiNonParametMd

Statistical inference in parametric models is generally well-understood, but parametric assumptions are unrealistic in many settings. Semiparametric and nonparametric models provide more flexible alternatives that may better reflect our knowledge of the problem at hand, but statistical inference in these models is often challenging. In this course, we will introduce the statistical theory and methods underlying targeted inference of Euclidean parameters in semiparametric and nonparametric models. We will begin by discussing aspects of semiparametric efficiency theory. We will then introduce several general-purpose methods of targeted estimation in these models. Finally, we will provide an overview of tools for analyzing the behavior of such estimators, emphasizing the role that modern machine learning methods can play. Throughout the course, we will illustrate these methods using problems from causal inference.

PreReq: STATISTC 607 & 608