Causality’24

This is the online page for the 2024 course on Causality and Causal Inference at FGV EMAp, ministered by Prof. Claudio Struchiner. I am the TA for the course, and will keep this page up-to-date with course information, planning and details.

Most of our classes will take on a “reading group style”: students will present selected papers and we will conduct group discussions on the topic.

Meetings

• 2024-03-12: Course details and introduction
  • Main references: A First Course in Causal Inference chapters 1, 2 and 10
• 2024-03-14: The propensity score, IPW and covariate shifts
  • Main references: A First Course in Causal Inference chapter 11
  • Notes on inference by sample reweighting: pdf
  • See also: A First Course in Causal Inference chapters 12 and 14
• 2024-03-19: Basic ML for CATE estimation: S-learners, T-learners and X-learners
  • Paper: Metalearners for estimating heterogeneous treatment effects using machine learning
• 2024-03-21: Overview of assumptions and fundamentals for causal inference
  • Paper: Disentangling causality: assumptions in causal discovery and inference
• 2024-03-26: More on Causal Graphs
  • Paper: What is a Causal Graph
• 2024-04-02 - 2024-04-09: Longitudinal causal inference
  • Main reference: Longitudinal data analysis, Chapter 23
  • See also: A First Course in Causal Inference chapter 29
• 2024-04-11: Overview of causal discovery
  • Paper: Methods and tools for causal discovery and causal inference
  • Paper: A Survey on Causal Discovery: Theory and Practice
• 2024-04-16: A look into contemporary causal discovery
  • Paper: Gradient-based neural DAG learning
• 2024-04-18: A look into causal inference bypassing causal discovery
  • Paper: Torwards Causal Inference with Minimal DAG Knowledge
• 2024-04-25: Causal conformal prediction
  • Background on conformal prediction:
    • A Gentle Introduction to Conformal Prediction and Distribution-Free Uncertainty Quantification
    • Conformal Prediction Under Covariate Shift
  • Paper: Conformal Inference of Counterfactuals and Individual Treatment Effects
• 2024-04-30: Sensitivity Analysis
  • Main references: A First Course in Causal Inference chapter 18 and Conformal Sensitivity Analysis for Individual Treatment Effects
  • See also: B-Learner: Quasi-Oracle Bounds on Heterogeneous Causal Effects Under Hidden Confounding
• 2024-05-02: Synthetic Controls
• 2024-05-09: Granger causality
  • Paper: Granger Causality: A Review and Recent Advances
  • See also: P. W. Holland - Statistics and Causal Inference, which features a discussion on the different paradigms of causality, including (briefly) Granger causality
• 2024-05-14: Causal Discovery Under Hidden Confounding
  • Paper: Nonlinear Causal Discovery with Latent Confounders
• 2024-05-16: Counterfactual Explanations
  • Paper: Shift-Robust Molecular Relational Learning with Causal Substructure
• 2024-05-21: Causal Inference with Agent-Based Models
  • Paper: Formalizing the Role of Agent-Based Modeling in Causal Inference and Epidemiology
• 2024-05-23: Data Fusion
  • Paper: Causal inference and the data-fusion problem
• 2024-05-28: Treatment Effect Estimation by Partial Identification
  • Paper: Approximate Causal Effect Identification under Weak Confounding
• 2024-06-04: Causal Discovery under Weaker Conditions
  • Paper: Localised Natural Causal Learning Algorithms for Weak Consistency Conditions
• 2024-06-06 - 2024-06-18: Causal Discovery and Estimation in Practice
• 2024-06-18: Software Packages for Causal Inference
  • Packages: DoWhy, EconML, CausalML
• 2024-06-20: Learning Theory for Causal Regression
  • Paper: Generalization Bounds for Causal Regression: Insights, Guarantees and Sensitivity Analysis
• 2024-06-25: Deep Learing for Treatment Effect Estimation — CEVAE
  • Paper: Causal Effect Inference with Deep Latent-Variable Models
• 2024-06-27: Mediation Analysis
  • Main references: A First Course in Causal Inference chapter 27 and Medflex
• 2024-07-02: Heterogeneity: Population Effects, Interaction and EFfect Modification
  • Main references: Heterogeneity’s Ruses: Some Surprising Effects of Selection on Population Dynamics, Effect modification, interaction and mediation: an overview of theoretical insights for clinical investigators, On the distinction between interaction and effect modification
• 2024-07-04: Targeted Learning
  • Main references: Targeted Learning: Causal Inference for Observational and Experimental Data chapters 1, 2, 3 and 4
• 2024-07-09: Tian-Pearl Bounds
  • Paper: Probabilities of causation: Bounds and identification
• 2024-07-11: Prediction-powered Generalization of Causal Inferences
  • Paper: Prediction-powered Generalization of Causal Inferences

Suggested Reading

• A First Course in Causal Inference: these are featureful lecture notes on Causal Inference by Peng Ding from UC Berkeley.
• Causal Inference for the Brave and True: this is a very friendly online material on causal inference that includes a number of modern topics.
• All of Statistics chapters 16 and 17

• D’ya like DAGs? A Survey on Structure Learning and Causal Discovery: this survey contains brief descriptions of many recently proposed methods for Causal Discovery.
• On the tractability of Causal Discovery:
  • On the Number of Experiments Sufficient and in the Worst Case Necessary to Identify All Causal Relations Among N Variables
  • The Hardness of Conditional Independence Testing and the Generalised Covariance Measure
• Conditional independence testing:
  • Model-X Knockoffs: Panning for Gold: Model-X Knockoffs for High-dimensional Controlled Variable Selection
  • Sequential e-values via knockoffs: Model-X Sequential Testing for Conditional Independence via Testing by Betting
  • GAN Knockoffs: Deep Knockoffs (2018)
• Causal ML:
  • R-learner: Quasi-Oracle Estimation of Heterogeneous Treatment Effects
  • B-learner: B-Learner: Quasi-Oracle Bounds on Heterogeneous Causal Effects Under Hidden Confounding
  • Double ML: Double/debiased machine learning for treatment and structural parameters
• On causal benchmarking:
  • Doing Great at Estimating CATE? On the Neglected Assumptions in Benchmark Comparisons of Treatment Effect Estimators
• Rashomon effect: Amazing Things Come From Having Many Good Models

This list will likely be updated over time. Additionally, most of our primary references will be in the form of papers to be read and discussed.