diff --git a/README.md b/README.md index 17bf545..054d7e5 100644 --- a/README.md +++ b/README.md @@ -1,6 +1,6 @@ # 『실무로 통하는 인과추론 with 파이썬』 -이 저장소는 한빛출판사에서 출간한 한국어판 "실무로 통하는 인과추론 with 파이썬"에서 참조하는 각종 자료 및 소스 코드와 예제 데이터를 담고 있습니다. 이 책은 다음 서점에서 절찬리에 판매중입니다([교보문고](https://product.kyobobook.co.kr/detail/S000212577153), 예스24[https://www.yes24.com/Product/Goods/125196916], [인터파크](https://book.interpark.com/product/BookDisplay.do?_method=detail&sc.shopNo=0000400000&sc.prdNo=356878902&sc.saNo=003002001&bid1=search&bid2=product&bid3=title&bid4=001), [알라딘](https://www.aladin.co.kr/shop/wproduct.aspx?ItemId=335177841), [한빛미디어](https://www.hanbit.co.kr/store/books/look.php?p_code=B6208936856)) +이 저장소는 한빛출판사에서 출간한 한국어판 "실무로 통하는 인과추론 with 파이썬"에서 참조하는 각종 자료 및 소스 코드와 예제 데이터를 담고 있습니다. 이 책은 다음 서점에서 절찬리에 판매중입니다([교보문고](https://product.kyobobook.co.kr/detail/S000212577153), [예스24](https://www.yes24.com/Product/Goods/125196916), [인터파크](https://book.interpark.com/product/BookDisplay.do?_method=detail&sc.shopNo=0000400000&sc.prdNo=356878902&sc.saNo=003002001&bid1=search&bid2=product&bid3=title&bid4=001), [알라딘](https://www.aladin.co.kr/shop/wproduct.aspx?ItemId=335177841), [한빛미디어](https://www.hanbit.co.kr/store/books/look.php?p_code=B6208936856)) ![img](./실무로%20통하는%20인과추론%20with%20파이썬/data/book_cover.jpg) @@ -17,12 +17,13 @@ 한국에서는 실무자가 접할 수 있는 인과추론에 대한 자료가 많이 부족하다고 생각합니다. 이 책이 여러분의 인과추론 학습에 있어 좋은 길잡이가 되길 바랍니다. +책에 대한 문의는 [제 블로그](https://jinsooshin.tistory.com/) 또는 [가짜연구소 Discord](https://discord.gg/HeHbFAvmSZ)를 통해 이야기해주세요. + 박지용 교수님의 ["인과추론과 데이터과학"](https://www.youtube.com/@causaldatascience) 강의와 함께 학습하면 더 좋습니다! # 역자 정보 ## [신진수](https://www.linkedin.com/in/jinsoo-shin-436060162/) -네오플을 거쳐 크래프톤의 데이터 분석가로 일하고 있습니다. 게임 업계에서 쌓은 커리어를 기반으로 <던전앤파이터>, <뉴스테이트 모바일>, <배틀그라운드 모바일> 등 다양한 장르의 게임에서 데이터 분석과 실험을 통해 유저 경험을 개선하는 데 기여했습니다. 비영리 데이터 사이언스 커뮤니티인 가짜연구소에서 인과추론팀을 운영 중입니다. 마테우스 파쿠레의 웹북 [「Causal Inference for The Brave and True」](https://github.com/CausalInferenceLab/Causal-Inference-with-Python)를 한국어로 번역하는 작업을 주도했습니다. +네오플을 거쳐 크래프톤의 데이터 분석가로 일하고 있습니다. 게임 업계에서 쌓은 커리어를 기반으로 <던전앤파이터>, <뉴스테이트 모바일>, <배틀그라운드 모바일> 등 다양한 장르의 게임에서 데이터 분석과 실험을 통해 유저 경험을 개선하는 데 기여했습니다. 비영리 데이터 사이언스 커뮤니티인 가짜연구소에서 인과추론팀을 운영 중입니다. 마테우스 파쿠레의 웹북 [「Causal Inference for The Brave and True」](https://github.com/CausalInferenceLab/Causal-Inference-with-Python)를 한국어로 번역하는 작업을 주도했습니다 ([블로그](https://jinsooshin.tistory.com/) / [Github](https://github.com/jsshin2022)) ## [가짜연구소 인과추론팀](https://pseudo-lab.com/6bbf03d9f11d4af687c0f03c6db39b1b) @@ -31,10 +32,28 @@ # 챕터별 자료 +### 참고 도서 +| 도서 | +| ------ | +| [Causal Inference Stanford STATS 361 강의 노트](https://web.stanford.edu/~swager/stats361.pdf) | +| Causal Inference for the Brave and True ([원서](https://github.com/matheusfacure/python-causality-handbook) / [번역서](https://github.com/CausalInferenceLab/Causal-Inference-with-Python)) | +| [Causal Inference for Statistics, Social, and Biomedical Sciences](https://www.amazon.com/Causal-Inference-Statistics-Biomedical-Sciences/dp/0521885884) | + + +### 1장-인과추론 소개 + +| 챕터 | 내용 | 자료 링크 | +| ------ |------|------| +| 1.6 인과효과 식별하기 | 더 알아보기 | [Causal Inference and Data Fusion in Econometrics](https://arxiv.org/abs/1912.09104) | + + ### 2장-무작위 실험 및 기초 통계 리뷰 | 챕터 | 내용 | 자료 링크 | | ------ |------|------| +| 2.2 A/B 테스트 사례 | 시뮬레이션 데이터와 실제 데이터 비교 | [A Randomized Assessment of Online Learning](https://www.aeaweb.org/articles?id=10.1257/aer.p20161057) | +| 2.4 가장 위험한 수식 | 하워드 웨이너의 유명한 글 | [The Most Dangerous Equation](https://www.americanscientist.org/article/the-most-dangerous-equation) | +| 2.5 추정값의 표준오차 | MIT의 통계학 입문 강좌 | [Introduction to statistics](https://www.youtube.com/watch?v=VPZD_aij8H0&list=PLUl4u3cNGP60uVBMaoNERc6knT_MgPKS0) | | 2.6 신뢰구간 | 실제 사례: 코로나19 백신의 효과 | [Efficacy and Safety of the mRNA-1273 SARS-CoV-2 Vaccine](https://www.nejm.org/doi/full/10.1056/nejmoa2035389) | | 2.8 p 값 | 실제 사례: 실제 사례: 대면 강의 vs. 온라인 강의 | [Is It Live or Is It Internet? Experimental Estimates of the Effects of Online Instruction on Student Learning](https://www.journals.uchicago.edu/doi/10.1086/669930) | | 2.10 표본 크기 계산 | 더 알아보기 | [A/B Testing Intuition Busters: Common Misunderstandings in Online Controlled Experiments](https://dl.acm.org/doi/abs/10.1145/3534678.3539160) | @@ -43,7 +62,68 @@ | 챕터 | 내용 | 자료 링크 | | ------ |------|------| | 3.7.2 랜덤화 재해석 | 민감도 분석과 부분 식별 | [Making Sense of Sensitivity: Extending Omitted Variable Bias](https://academic.oup.com/jrsssb/article/82/1/39/7056023) | -| 3.8.1 충돌부를 조건부 설정 | 더 알아보기 | [DAGitty](https://www.dagitty.net/) | -| 3.8.1 충돌부를 조건부 설정 | 더 알아보기 | [A Crash Course in Good and Bad Controls](https://journals.sagepub.com/doi/epub/10.1177/00491241221099552) | +| 3.8.1 충돌부를 조건부 설정 | 더 알아보기 | [DAGitty](https://www.dagitty.net/), [A Crash Course in Good and Bad Controls](https://journals.sagepub.com/doi/epub/10.1177/00491241221099552) | + +### 4장-유용한 선형회귀 +| 챕터 | 내용 | 자료 링크 | +| ------ |------|------| +| 4.1 선형회귀의 필요성 | OLS 연구 | [Difference-in-differences with variation in treatment timing](https://www.sciencedirect.com/science/article/abs/pii/S0304407621001445), [Interpreting Ols Estimands When Treatment Effects are Heterogeneous: Smaller Groups](https://direct.mit.edu/rest/article/104/3/501/97692/Interpreting-OLS-Estimands-When-Treatment-Effects), [Contamination Bias in Linear Regressions](https://arxiv.org/abs/2106.05024) | +| 4.7.5 평균 제거와 고정효과 | 패널데이터를 사용한 인과추론 논문 | [On the Pooling of Time Series and Cross Section Data](https://www.jstor.org/stable/1913646) | +| 4.9 중립 통제변수 | 잡음 제거 기법 | [Improving the Sensitivity of Online Controlled Experiments by Utilizing Pre-Experiment Data](http://robotics.stanford.edu/~ronnyk/2013-02CUPEDImprovingSensitivityOfControlledExperiments.pdf) | + + +### 5장-성향점수 +| 챕터 | 내용 | 자료 링크 | +| ------ |------|------| +| 5.1 관리자 교육의 효과 | 민감도 분석과 부분 식별 | [Estimating Treatment Effects with Causal Forests: An Application](https://arxiv.org/abs/1902.07409) | +| 5.3.3 성향점수 매칭 | 민감도 분석과 부분 식별 | [Why Propensity Scores Should Not Be Used for Matching](https://gking.harvard.edu/publications/why-Propensity-Scores-Should-Not-Be-Used-Formatching) | +| 5.3.5 역확률 가중치의 분산 | 실제 사례: 인과적 콘텍스트 밴딧 | [A Contextual-Bandit Approach to Personalized News Article Recommendation](https://arxiv.org/abs/1003.0146) | +| 5.5 이중 강건 추정 | 이중 강건 추정량 주석 | [An Introduction to the Augmented Inverse Propensity Weighted Estimator](https://www.law.berkeley.edu/files/AIPW(1).pdf) | +| 5.5.2 결과 모델링이 쉬운 경우 | 더 알아보기 | [Comment: Performance of Double-Robust Estimators When “Inverse Probability” Weights Are Highly Variable](https://arxiv.org/pdf/0804.2965.pdf) | +| 5.6 연속형 처치에서의 일반화 성향점수 | 연속형 처치 | [Causal inference with a continuous treatment and outcome: Alternative estimators for parametric dose-response functions with applications](https://www.proquest.com/openview/888e50198b2563222f8e2030e5a8c9e7/1?pq-origsite=gscholar&cbl=18750) | + -### 4장-유용한 선형회귀 \ No newline at end of file +### 6장-이질적 처치효과 +| 챕터 | 내용 | 자료 링크 | +| ------ |------|------| +| 6.7 누적 이득 곡선 | 더 알아보기 | [Nubank fklearn](https://github.com/nubank/fklearn), [Causal inference and uplift modeling a review of the literature](https://proceedings.mlr.press/v67/gutierrez17a/gutierrez17a.pdf), [Empirical Analysis of Model Selection for Heterogeneous Causal Effect Estimation](https://arxiv.org/abs/2211.01939) | +| 6.8 목표 변환 | 더 알아보기 | [Intelligent Credit Limit Management in Consumer Loans Based on Causal Inference](https://arxiv.org/abs/2007.05188) | +| 6.9.2 이진 결과 | 더 알아보기 | [Causal Classification: Treatment Effect Estimation vs. Outcome Prediction](https://jmlr.org/papers/v23/19-480.html) | + + +### 7장-메타러너 +| 챕터 | 내용 | 자료 링크 | +| ------ |------|------| +| 7.1 이산형 처치 메타러너 | 인과추론 라이브러리 | [EconML 공식문서](https://econml.azurewebsites.net/), [CausalML 공식문서](https://causalml.readthedocs.io/en/latest/about.html) | +| 7.1.1 T 러너 | 더 알아보기 | [Metalearners for estimating heterogeneous treatment effects using machine learning](https://www.pnas.org/doi/full/10.1073/pnas.1804597116) | +| 7.2.1 S 러너 | S 러너의 편향 | [Double/Debiased/Neyman Machine Learning of Treatment Effects](https://www.aeaweb.org/articles?id=10.1257/aer.p20171038) | +| 7.2.2 이중/편향 제거 머신러닝 | 트리 기반 및 신경망 러너 | [Nonparametric Estimation of Heterogeneous Treatment Effects:From Theory to Learning Algorithms](https://proceedings.mlr.press/v130/curth21a/curth21a.pdf), [Learning Representations for Counterfactual Inference](https://proceedings.mlr.press/v48/johansson16.pdf) | + + +### 8장-이중차분법 +| 챕터 | 내용 | 자료 링크 | +| ------ |------|------| +| 8.2 표준 이중차분법 | 실제 사례: 최저임금과 고용 | [Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jersey and Pennsylvania](https://davidcard.berkeley.edu/papers/njmin-aer.pdf) | +| 8.2.5 추론 | 더 알아보기 | [When Should You Adjust Standard Errors for Clustering?](https://academic.oup.com/qje/article/138/1/1/6750017), [An Exact and Robust Conformal Inference Method for Counterfactual and Synthetic Controls](https://www.tandfonline.com/doi/full/10.1080/01621459.2021.1920957) | +| 8.3 식별 가정 | 더 알아보기 | [When Is Parallel Trends Sensitive to Functional Form?](https://cowles.yale.edu/sites/default/files/2022-11/2010.04814.pdf) | +| 8.3.2 비기대 가정과 SUTVA | 공간적 파급 효과 | [Difference-in-Differences Estimation with Spatial Spillovers](https://arxiv.org/abs/2105.03737) | +| 8.3.5 피드백 없음 | 순차적 무시 가능성 | [Causal Inference with Time-Series Cross-Sectional Data: A Reflection](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3979613) | +| 8.3.6 이월 효과와 시차종속변수 없음 | 순차적 무시 가능성 | [When Should We Use Unit Fixed Effects Regression Models for Causal Inference with Longitudinal Data?](https://imai.fas.harvard.edu/research/files/FEmatch.pdf) | +| 8.6.3 최종 결과 | 이중 강건 이중차분법 | [Doubly robust difference-in-differences estimators](https://www.sciencedirect.com/science/article/abs/pii/S0304407620301901) | +| 8.7 처치의 시차 도입 | 이중 강건 이중차분법 | [Difference-in-differences with variation in treatment timing](https://www.sciencedirect.com/science/article/abs/pii/S0304407621001445) | +| 8.7 처치의 시차 도입 | 실제 사례: 개발도상국의 고등교육과 성장 | [Higher Education Expansion, Labor Market, and Firm Productivity in Vietnam](https://sites.google.com/view/khoavu-umn/job-market-paper) | +| 8.7.1 시간에 따른 이질적 효과 | 더 알아보기 | [Estimating dynamic treatment effects in event studies with heterogeneous treatment effects](https://www.sciencedirect.com/science/article/abs/pii/S030440762030378X), [Two-Way Fixed Effects, the Two-Way Mundlak Regression, and Difference-in-Differences Estimators](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3906345), [Difference-in-Differences with multiple time periods](https://www.sciencedirect.com/science/article/abs/pii/S0304407620303948) | + + +### 9장-통제집단합성법 +| 챕터 | 내용 | 자료 링크 | +| ------ |------|------| +| 9.4 표준 통제집단합성법 | 통제집단합성법에 대한 가정 | [Using Synthetic Controls: Feasibility, Data Requirements, and Methodological Aspects](https://www.aeaweb.org/articles?id=10.1257/jel.20191450) | +| 9.7 추론 | 더 알아보기 | [An Exact and Robust Conformal Inference Method for Counterfactual and Synthetic Controls](https://arxiv.org/abs/1712.09089) | +| 9.8.4 통제집단합성법과 이중차분법 | 원본 합성 이중차분법 | [Synthetic Difference-in-Differences](https://www.aeaweb.org/articles?id=10.1257/aer.20190159) | +| 9.9 요약 | 일반화 통제집단합성법 | [Generalized Synthetic Control Method: Causal Inference with Interactive Fixed Effects Models](https://www.cambridge.org/core/journals/political-analysis/article/generalized-synthetic-control-method-causal-inference-with-interactive-fixed-effects-models/B63A8BD7C239DD4141C67DA10CD0E4F3), [A Bayesian Alternative to Synthetic Control for Comparative Case Studies](https://www.cambridge.org/core/journals/political-analysis/article/bayesian-alternative-to-synthetic-control-for-comparative-case-studies/C23BD67E4BBBB8C88ADAEAE169696A45) | +| 9.9 요약 | 실제 사례: causalimpact 라이브러리 | [CausalImpact Github](https://github.com/google/CausalImpact?tab=readme-ov-file) | + +### 10장-스위치백 실험 +| 챕터 | 내용 | 자료 링크 | +| ------ |------|------| \ No newline at end of file