Applied Mathematics and Statistics presents a Wierman Lecture Series w/Cory Zigler

Location: Gilman 132

When: May 11th at 1:30 p.m.

Title: Causal Inference in Air Quality Regulation: An Overview and Two Topics in Statistics and Machine Learning

Abstract: Methods for causal inference have emerged at the forefront of scientific debates about air quality regulation in the United States and elsewhere. This outlines the increasing relevance of causal inference in this domain and outlines two modern topics in statistical and machine learning research tailored to investigation of environmental regulations. The first is bipartite causal inference with interference, which arises when evaluating causal effects of treatments applied at point sources of air pollution in a manner that reflects the nature of long-range pollution transport. We briefly outline methodology that amounts to making causal inferences on a network governed by the mechanics of how air pollution is formed and transported through the atmosphere. The second is representation learning methods for nonlocal confounding, motivated by the ubiquitous problem of parsing regulation-induced changes in air pollution from meteorology. The proposed methodology adopts a U-net architecture to represent local and regional weather patterns that can be input into strategies for nonlocal confounding adjustment. Both methodologies are illustrated through investigating the impacts of regulations targeting power plants.

Zoom Link: https://wse.zoom.us/j/95738965246