Introduction to Causal Inference

İzlencenin pdf versiyonu burada. Eğer seminere katılmak isterseniz [email protected]'a ilginizi açıklayan kısa bir paragrafı içeren bir e-posta atmanızı rica ediyorum. İlk sürümü ufak bir grupla yürütmek arzusundayım. Bu nedenle, eklediğiniz paragraflara başvurarak bir seçim yapmak durumunda kalabilirim.

instructor: direnç erşahin
meetings: online via zoom. Monday evenings - 19:00 to 21:45 Turkey time (CET 17:00 to 19:45)
prerequisites: no distinct prerequisites other than a genuine interest on the subject.

i.aim

Causality has long been a primary issue of interest for both philosophy and science. Recent developments on the topic paved the way for a wider utilisation of causal inference methods in various fields, including machine learning, economics, and epidemiology. This lecture formation intends to provide an introduction to the basics of causality. By implementing and interpreting causal inference methods, we will try to develop an interdisciplinary and practical understanding of causal analysis.

keywords: Causality. Causal inference. Potential outcomes. Bayesian Networks. Structural Causal Models. Interventions. Do-calculus. Counterfactuals.

ii. content

This is a less quantitative, seminar version of a full-semester course. The 10-week lecture series is composed of three modules. These modules address the same issue at different levels: why we do what we do, when we are deriving a causal inference. In this context, we will employ why and how questions to probe available causal inference methods.
If you would like to take a bite of the seminar content, Kevin Hartnett's exciting Q&A with Judea Pearl might be much helpful: https://www.quantamagazine.org/to-build-truly-intelligent-machines-teach-them-cause-and-effect-20180515/

iii. course plan

The meetings will be held online, via zoom. Discussions will be primarily in Turkish. However, since all the material is in English, fluency in both languages is required. Judea Pearl and Dana Mackenzie's highly influential work, The Book of Why, will be our main text. For most of the weeks, we will read a chapter of this book. Some academic texts, articles and book sections, will accompany these chapters and help us to ground our discussion. I will provide the readings on a weekly basis. I will also offer a number of weekly assignments to implement our findings. We will start on February 6, 2023.

iv. schedule

Week ##	Meeting date	Readings
Module I - 3 weeks
Week 01	06 February	Model-building. Causality. [01] Williamson, T. (2018). Model-Building. [02] Illari, P., & Russo, F. (2014). Prelude to Causality. [03] Beebee, H. (2014). Causation.
Week 02	13 February	Probability: Frequentism vs. Bayesianism. Statistical inference. Causal inference. [01] Neal, B. (2020). Motivation: Why You Might Care. [02] Koller, D. & Friedman, N. (2009). Probability Theory. [03] Pearl, J., Glymour, M., & Jewell, N. P. (2016). Probability and Statistics.
Week 03	20 February	Graph Theory. Bayesian Networks. Directed acyclic graphs. [01] Pearl, J. (2018). The Ladder of Causation. [02] Darwische, A. (2010). Bayesian Networks.
Module II - 4 weeks
Week 04	27 February	Causal Models. Reichenbach's common cause principle. Selection bias. [01] Pearl, J. (2018). From Evidence to Causes: Reverend Bayes Meets Mr. Holmes. [02] Neal, B. (2020). The Flow of Association and Causation in Graphs.
Week 05	06 March	Confounding. d-separation. [01] Pearl, J. (2018). Confounding and Deconfounding: Or, Slaying the Lurking Variable. [02] Pearl, J. (2009). The d-Separation Criterion.
Week 06	13 March	Randomized trials. Observational studies. Backdoor path criterion. [01] Pearl, J. (2018). The Skillful Interrogation of Nature: Why RCTS Work. [02] Neal, B. (2020). Randomized Experiments.
Week 07	20 March	Interventions. Front-door adjustment. Controls. [01] Pearl, J. (2018). Beyond Adjustment: The Conquest of Mount Intervention [02] Cinelli, C., Forney, A., & Pearl, J. (2021). A Crash Course in Good and Bad Controls.
Module III - 3 weeks
Week 08	27 March	Do-Calculus. [01] Pearl, J. (2019). On the Interpretation of do(x).
Week 09	03 April	Counterfactuals. [01] Pearl, J. (2018). Counterfactuals: Mining Worlds That Could Have Been. [02] Williamson, T., (2009). Knowledge of Counterfactuals.
Week 10	10 April	Final remarks. Wrapping things up. [01] Pearl, J. (2019). The 7 Tools of Causal Inference with Reflections on Machine Learning.