因果推論
Causal Inference
| 節 | 週二 |
|---|---|
5 13:20–14:10 | 因果推論 A406 3 節連堂 |
6 14:20–15:10 | |
7 15:30–16:20 |
* 根據陽明交大上課時間表所列
Course description: This course (IST5556: causal inference and mediation analysis) introduce methods for causal inference, mediation analysis and interaction analysis. The course begins with presenting the three basic techniques widely used in causal theory development: (1) Counterfactual (potential) outcome model, (2) Direct Acyclic Graphs, and (3) Sufficient Component Cause model; and then explored in specific topics such as time-varying system, mediation analysis, interaction analysis, and different methods for estimation. Students who take this course are expected to be equipped with sufficient techniques to develop causal theory in a certain topic and conduct data analysis based on causal model. Course Objectives: 1. Understand the framework of studies in field of causal theory, mediation analysis and interaction analysis. 2. Formulate and address causal problems based on counterfactual models, causal diagram, and sufficient component cause model 3. Estimate causal effect and conduct mediation analysis using various modeling approaches 4. Identify the advantages and disadvantages of causal inference method and traditional statistical analysis.
No prerequisites for this course, but familiarity with basic statistical inference and regression model will be assumed. Some exposure to data analysis with statistics software (especially R) will be helpful.
The course grade will be based on midterm exam (40%), class participation (20%), and final project presentation and take-home final exam (40%). Students are encourage to discuss and work on the homework assignment in groups. Exams are in-class and close book, almost based from the homework assignments. Two pages of cheat sheet are acceptable.
| 週次 | 主題 |
|---|---|
| 第 1 週 | Introduction Topic 0. How to read/write a paper of causal inference |
| 第 2 週 | Topic 1. Basic definition and measurement for Epidemiology |
| 第 3 週 | Topic 2. Visualization of causal structure: Direct Acyclic Graphs (DAGs) |
| 第 4 週 | Topic 3. From association to causality – counterfactual model |
| 第 5 週 | Topic 4. From God’s Table to Your Table: identification and causal assumptions |
| 第 6 週 | Topic 5. Estimation and modeling (1) |
| 第 7 週 | Topic 5. Estimation and modeling (2) |
| 第 8 週 | Spring Break |
| 第 9 週 | Topic 6. Concept and definition of causal effects for mediation analysis |
| 第 10 週 | Review for Midterm Exam |
| 第 11 週 | Midterm Exam |
| 第 12 週 | Software Practice |
| 第 13 週 | Topic 7. Identification of causal effects for mediation analysis |
| 第 14 週 | Topic 8. Estimation of causal effects for mediation analysis |
| 第 15 週 | Final project presentation (1、2): Causal model v.s. Statistical model 、Association v.s. Causality |
| 第 16 週 | Final project presentation (3): Counterfactual |
| 第 17 週 | Final project presentation (4、5): Causal inference v.s. AI 、Unification of causal inference and machine learning – Targeted maximum likelihood estimation (TMLE) Course Feedback |
| 第 18 週 | Final exam (Take home) |
Required reading: 1. Hernán MA, Robins JM (2019). Causal Inference. Boca Raton: Chapman & Hall/CRC, forthcoming. It can be freely downloaded from the following website: https://www.hsph.harvard.edu/miguel-hernan/causal-inference-book/ (Chapters 1~3, 6, and 7) 2. VanderWeele T. Explanation in Causal Inference: Methods for Mediation and Interaction: Oxford University Press; 2015. (Chapters 1, 2) 3. 因果革命:人工智慧的大未來 (The Book of Why: The New Science of Cause and Effect)。 作者: 朱迪亞・珀爾, 達納・麥肯錫(Judea Pearl, Dana Mackenzie)。行路出版社。 4. VanderWeele, Tyler J., and Stijn Vansteelandt. "Conceptual issues concerning mediation, interventions and composition." Statistics and its Interface 2.4 (2009): 457-468. 5. VanderWeele, Tyler J., and Stijn Vansteelandt. "Odds ratios for mediation analysis for a dichotomous outcome." American journal of epidemiology 172.12 (2010): 1339-1348. 6. VanderWeele, Tyler J., Stijn Vansteelandt, and James M. Robins. "Effect decomposition in the presence of an exposure-induced mediator-outcome confounder." Epidemiology (Cambridge, Mass.) 25.2 (2014): 300. Srongly recommended reading: 1. Lin, Sheng-Hsuan, and Tyler VanderWeele. "Interventional Approach for Path-Specific Effects." Journal of Causal Inference 5.1 (2017). 2. Daniel, R. M., et al. "Causal mediation analysis with multiple mediators." Biometrics 71.1 (2015): 1-14. Reference: 1. Rothman, K.J., Greenland, S. and Lash, T.L. (2008). Modern Epidemiology, 3rd edition. Philadelphia: Lippincott Williams and Wilkins. 2. Pearl, J. (2000). Causality: Models, Reasoning, and Inference. Cambridge: Cambridge University Press. 3. MacKinnon DP. Introduction to statistical mediation analysis: Routledge; 2008.
- 地點
- Room 413, Assembly Building I
- 時間
- Tuesday by appointment. Please inform the instructor one day in advance if you would like to use the office hour.
- 聯絡方式
- shenglin@stat.nctu.edu.tw (03-5712121 ext: 56822)