Academics
  /  
Courses
  /  
Descriptions
COMP_SCI 374: Causal Graphical Models


VIEW ALL COURSE TIMES AND SESSIONS

Prerequisites

Permission of Instructor

Description

We know that correlation does not imply causation, but careful analyses of correlations are often our only way to quantify cause and effect in domains ranging from healthcare to education. This courses introduces causal inference methods, primarily using probabilistic graphical models, to identify and estimate counterfactual quantities as functions of observational data. We will discuss common challenges to causal inference, including confounding bias, missing data, measurement error, and selection bias. The final project will allow students to choose a dataset on which to perform a causal data analysis.

  • Formerly Comp_Sci 396 Modeling Relationships with Causal Inference - last offer was Spring 2024
  • This course fulfills Technical Elective area.

REFERENCE TEXTBOOKS: N/A
REQUIRED TEXTBOOK: Causal Inference: What If by Miguel Hernán and James M. Robins (2020)
https://www.hsph.harvard.edu/miguel-hernan/causal-inference-book/

COURSE COORDINATORS: Zach Wood-Doughty

COURSE INSTRUCTOR : Prof. Zach Wood-Doughty

COURSE GOALS: By the end of the course, each student should be comfortable:
 - Explaining how different sources of bias can invalidate causal inferences
 - Creating a graphical model that represents a given dataset
 - Identifying whether a causal effect is identified in a given graphical model
 - Critiquing published papers and articles that make causal claims