Causal Foundations for Pragmatic Data Science
Name
squires-csquires-phd-eecs-2025-thesis.pdf
Description
Thesis PDF
Size
9.42 MB
Format
Adobe PDF
Checksum (MD5)
a33bf1eb63be243363c72a4db774b178
Author(s)
Squires, Chandler
Advisor(s)
Uhler, Caroline
Sontag, David
Date Issued
February 2025
Publisher
Massachusetts Institute of Technology
Abstract
A key goal of scientific discovery is the acquisition of knowledge that is practically useful for societal endeavors, such as the development of medicine or the design of fruitful economic policies. In this thesis, I place front and center the role that scientific models play in the process of decision-making, emphasizing the importance of causal models in science, i.e., models which describe the possible effects of actions upon a system. The work contained explores central topics in this domain, including causal discovery (learning causal models from data), causal representation learning (learning how to coarse-grain observations into causally sensible “macro-variables”), and end-to-end causal inference (the interplay between causal discovery and downstream decision-making).
MIT Department
Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science
Terms of Use
In Copyright - Educational Use Permitted
Copyright retained by author(s)
Persistent DSpace Link
