Brady Neal - Causal Inference
- My summary
- Introduction to Causal Inference
- Potential Outcomes
- The Flow of Causation and Association in Graph
- Causal Models
- Randomized Experiments and Identification
- Estimation
- Unobserved Confounding, Bounds, and Sensitivity Analysis
- Instrumental Variables
- Difference-in-Difference
- Causal Discovery from Observational Data
- Causal Discovery from Interventions
- Transfer Learning and Transportability
- Counterfactuals and Mediation
Korea Summer Session on Causal Inference 2021
- git wiki에 간단히 정리
- 01 인과추론의 다양한 접근법, Potential Outcome Framework, 인과적 사고방식
- 02 인과추론을 위한 연구 디자인, RCT, Quasi-Experiment, DID & Regression Discontinuity
- 05 준실험 연구 사례 2: 스마트 스티커가 컨텐츠 소비에 미치는 영향
- 07 인과 그래프, 인과그래프에서 변수통제, 인과그래프에서의 인과추론 전략, 인과 그래프의 응용
- 11 인과추론과 예측방법론의 차이, 실증연구에서의 빅데이터와 머신러닝의 역할, 인과추론에서의 머신러닝의 활용, 인과추론 기반의 예측 모델링 평가
- 13 머신러닝의 해석 가능성과 인과추론, 인과추론을 위한 머신러닝 모델
- 14 신약 개발에서의 인과추론의 역할과 한계, 머신러닝을 활용한 heterogeneity in Treatment effect
Causal Inference for the Brave and True
- My summary
- 01 Introduction to Causality
- 02 Randomized Experiments
- 03 Stats review
- 04 Graphical Causal Models
- 05 The Unreasonable Effectiveness of Linear Regression
- 06 Grouped and Dummy Regression
- 07 Beyond Confounders
- 08 Instrumental Variables
- 09 Non Compliance and LATE
- 10 Matching
- 11 Propensity Score
- 12 Doubly Robust Estimaion
- 13 Difference In Difference
- 14 Panel Data and Fixed Effects
- 15 Synthetic Control
- 16 Regression Discontinuity Design
- 18 Heterogeneous Treatment Effects and Personalization
- 19 Evaluation Causal Models
- 20 Plug and Play Estimators
Korea Summer Workshop on Causal Inference 2022 (Bootcamp시리즈)
- Bootcamp 1: 인과추론과 예측 방법론의 차이, 인과추론의 어려움과 인과추론 전략
- Bootcamp 2: 잠재적결과 프레임워크, 무작위 통제실험, 인과추론 관점에서의 회귀분석, 매칭과 역확률가중치
- Bootcamp 3: 디자인 기반의 인과추론, 준실험 분석방법론, 이중차분법, 가상의 통제집단
- Bootcamp 4: 도구변수, 인과추론 관점에서의 도구변수, 회귀 불연속, 통제함수와 선택모형
- Bootcamp 5: 인과 그래프, 디자인 기반의 인과추론에서의 인과 그래프 활용, 구조적 인과모형
- Industry, 프로덕트 애널리틱스에서의 인과추론의 활용 사례
- Health Informatics, 머신러닝을 활용한 이질적 인과효과 추정
Causal Inference and Discovery in Python
- 책에 있는 실습 코드 따라하기
- 00 dowhy 사용법
- 01 match (dowhy)
- 02 metalearner (dowhy, econml)
- 03 DML (dowhy, econml)
- 04 CausalForest (dowhy, econml)
- 05 uplift (econml)
- 06 synthetic control (causalpy)
- 07 causal discovery (gCastle)
Heterogeneous treatment effect estimation, uplift
- Causal Inference and Uplift Modeling A review of the literature, 2016
review
- Double machine learning for treatment and causal parameters, 2016
- Metalearners for estimation heterogeneous treatment effects using machine learning, 2019
- Estimation and Inference of Heterogeneous Treatment Effects using Random Forests, 2018
- (to read, balanced representation learning) Estimation individual treatment effect: generalization bounds and algorithms (2018)
causal inference
- (to read) Selection on Observed and Unobserved Variables: Assessing the Effectiveness of Catholic Schools
- observable confounder만 사용하는 방법 (regression, matching, weighting)의 경우 검증을 하는게 좋다
- blog
- DoWhy tutorial
- Heterogeneous Treatment Effect Estimation tutorial
- metalearner (T-learner)
- metalearner (X-learner)