Part I The Causal Inference Problems

1.1. The Philosophy of Causal Inference.

Referring to Oxford English Dictionary, etiology, a word derived from the Greek αἰτιολογία (aitiología) “giving a reason for” (αἰτία, aitía, "cause"); and -λογία (-logía) is the study of causes, origins, or reasons behind the way that things are. Causal inference is stronger than the inference of association in analyzing the response of an effect variable when the cause of the effect variable is changed.

A case study: Card, David. "The causal effect of education on earnings." Handbook of labor economics 3 (1999): 1801-1863. https://doi.org/10.1111/j.1468-0084.2012.00708.x

Related readings:

• Falcon, Andrea, "Aristotle on Causality", The Stanford Encyclopedia of Philosophy (Spring 2022 Edition), Edward N. Zalta (ed.), URL = <https://plato.stanford.edu/archives/spr2022/entries/aristotle-causality/>

• Gallow, J. Dmitri, "The Metaphysics of Causation", The Stanford Encyclopedia of Philosophy (Fall 2022 Edition), Edward N. Zalta & Uri Nodelman (eds.), URL = <https://plato.stanford.edu/archives/fall2022/entries/causation-metaphysics/>.

• Frisch, Mathias, "Causation in Physics", The Stanford Encyclopedia of Philosophy (Spring 2022 Edition), Edward N. Zalta (ed.), URL = <https://plato.stanford.edu/archives/spr2022/entries/causation-physics/>.

1.2. Causal Inference in Nobel Prize of Economists.

• Lab Experiments 2002:

  • Press Release:

    https://www.nobelprize.org/prizes/economic-sciences/2002/press-release/

  • The Nobel Award Ceremony Video: https://www.nobelprize.org/prizes/economic-sciences/2002/award-video/

  • The Nobel Award Ceremony Speech: https://www.nobelprize.org/prizes/economic-sciences/2002/ceremony-speech/

  • The Nobel Banquet Video:

    https://www.nobelprize.org/prizes/economic-sciences/2002/banquet-video/

    • Daniel Kaheneman: “for having integrated insights from psychological research into economic science, especially concerning human judgment and decision-making under uncertainty.”

    • Vernon L. Smith: “for having established laboratory experiments as a tool in empirical economic analysis, especially in the study of alternative market mechanisms.”

• Field Experiments 2019:

  • Press release:

    https://www.nobelprize.org/prizes/economic-sciences/2019/press-release/

  • Nobel Prize Award Ceremony Video: https://www.nobelprize.org/prizes/economic-sciences/2019/award-video/

  • The Nobel Prize Award Ceremony Speech:

    https://www.nobelprize.org/prizes/economic-sciences/2019/ceremony-speech/

    • Abhijit Banerjee, Esther Duflo, and Michael Kremer “for their experimental approach to alleviating global poverty.”

• Natural Experiments 2021:

  • The press release:

    https://www.nobelprize.org/prizes/economic-sciences/2021/press-release/

  • The Nobel Prize Award Ceremony: https://www.nobelprize.org/prizes/economic-sciences/2021/award-video/

  • The Nobel Prize Award Ceremony Speech:

    https://www.nobelprize.org/prizes/economic-sciences/2021/ceremony-speech/

    • David Card: “for his empirical contributions to labor economics.”

    • Joshua D. Angrist and Guido W. Imbens “for their methodological contributions to the analysis of causal relationships.”

• The course materials of Prof. Susan Athey (Duke Alumni and the first female winner of the Clark Medal) and Prof. Guido W. Imbens (Nobel Prize Winner this year) at Stanford University.

  • Course resources:

    https://ml4econ.github.io/course-spring2019/resources.html

  • Course Syllabus: https://athey.people.stanford.edu/sites/g/files/sbiybj5686/f/phdmlsyllabus.pdf

  • Research papers:

    https://athey.people.stanford.edu/research

  • Open Educational Resources Course Videos [Machine Learning & Causal Inference: A Short Course].

1.3. Causal Inference in Turing Awards of Computer Scientist.

• Judea Pearl 2011: “for fundamental contributions to artificial intelligence through the development of a calculus for probabilistic and causal reasoning.”

  https://amturing.acm.org/award_winners/pearl_2658896.cfm

  • The Book of Why: the new science of cause and effect: https://www.amazon.com/Book-Why-Science-Cause-Effect/dp/046509760X

  • The paper published on Sociological Methods and Research (SMR):

    Cinelli, C., Forney, A., & Pearl, J. (2022). A Crash Course in Good and Bad Controls. Sociological Methods & Research, 0(0). https://doi.org/10.1177/00491241221099552

  • NeurIPS 2021 Workshop “Causal Inference & Machine Learning: Why now?”

    https://nips.cc/Conferences/2021/Schedule?showEvent=21871

• Yoshua Bengio 2018: “for conceptual and engineering breakthroughs that have made deep neural networks a critical component of computing”

  https://amturing.acm.org/award_winners/bengio_3406375.cfm

  • ICML 2022 Tutorial Causality and Deep Learning: Synergies, Challenges and the Future

    https://sites.google.com/view/causalityanddeeplearning/start

  • Towards Causal Representation Learning: An AI & Deep Learning Perspective on Causality:

    https://arxiv.org/abs/2102.11107

  • NeurIPS 2021 “The Causal-Neural Connection: Expressiveness, Learnability, and Inference”:

    • https://papers.nips.cc/paper/2021/hash/5989add1703e4b0480f75e2390739f34-Abstract.html

    • https://neurips.cc/virtual/2021/workshop/21871

  • Introduction to Causal Inference by Yoshua Bengio’s student Brady Neal

    https://www.bradyneal.com/causal-inference-course

  • New In ML at NeurIPS 2022: A Deep Learning Journey

    https://nehzux.github.io/NewInML2022NeurIPS/

Part II Other Open Educational Resources

• Microsoft Research Summit 2021: Causal Machine Learning

  • Website [Video Recordings]:

    https://www.microsoft.com/en-us/research/videos/causal-machine-learning/

• Microsoft Research at NeurIPS

  • Competition: Causal Insights for Learning Paths in Education

    https://eedi.com/projects/neurips-2022

  • NeurIPS Competition Instructions and Guide: Causal Insights for Learning Paths in Education

    https://arxiv.org/abs/2208.12610

    Deep End-to-end Causal Inference

    https://openreview.net/forum?id=6DPVXzjnbDK

  • Rhino: Deep Causal Temporal Relationship Learning with history-dependent noise 

    https://openreview.net/forum?id=Z53CEX9jh4E

  • Causal Reasoning in the Presence of Latent Confounders via Neural ADMG Learning

    https://openreview.net/forum?id=RQQxCLpCVr9

  • A Causal AI Suite for Decision-Making

    https://openreview.net/forum?id=-gVJ1_lD1RH

Part III The Algorithms and Workflow for Causal Inference

1. Regression Discontinuity

• Literature: https://www.sciencedirect.com/topics/economics-econometrics-and-finance/regression-discontinuity-design

• Practical Guide:

  • https://mixtape.scunning.com/06-regression_discontinuity

  • https://www.mdrc.org/publication/practical-guide-regression-discontinuity

• Collections: https://rdpackages.github.io/ (Python package not ready yet)

• Sample Code: https://colab.research.google.com/github/scunning1975/mixtape_learnr/blob/main/python/Regression_Discontinuity.ipynb#scrollTo=gdYEAGvwoGR

  • What is your Y (effect) variable

  • When is the cut-off event (treatment variable I)?

  • How frequent is your observation for capturing time trends (variable X)?

  • What are the controls (variable Z)?

  • What are the confounding factors not considered in I, X, and Z?

• DoWhy (Microsoft Research): provides multiple identification and validation methods

  • GitHub: https://github.com/py-why/dowhy

  • Webpage: https://www.microsoft.com/en-us/research/project/dowhy/

  • Paper: https://arxiv.org/abs/2011.04216

• EconML (Microsoft Research): provides multiple causal estimation methods

  • GitHub: https://github.com/microsoft/EconML

  • Webpage: https://www.microsoft.com/en-us/research/group/alice/

  • Paper: https://onlinelibrary.wiley.com/doi/full/10.3982/ECTA15732

• DECI (Microsoft Research): provides a framework for end-to-end causal inference which can also be used for discovery or estimation alone.

  • GitHub: https://github.com/microsoft/project-azua; https://github.com/microsoft/causica

  • Paper: https://arxiv.org/abs/2202.02195

  • Starter Datasets: https://github.com/microsoft/causica

  • Starter Codebook: https://github.com/microsoft/causica/tree/main/examples

• ShowWhy (Microsoft Research): provides a non-code end-to-end analysis in a user-friendly Graphical User Interface (GUI)

  • GitHub: https://github.com/microsoft/showwhy

• Alex and Causal Python

  • The book on Amazon: https://www.amazon.com/Causal-Inference-Discovery-Python-learning/dp/1804612987

  • The recommended casualDisco for checking assumptions: https://shiny.sund.ku.dk/zms499/causalDisco/

• gCastle (Huawei): for identifying causal graphs.

  • GitHub: https://github.com/huawei-noah/trustworthyAI/tree/master/gcastle

  • Blog: https://towardsdatascience.com/beyond-the-basics-level-up-your-causal-discovery-skills-in-python-now-2023-cabe0b938715?gi=960bfa4ade43

  • Paper: https://arxiv.org/abs/2111.15155

• CausalPy (PyMC Labs): for synthetic control and regression discontinuity, especially in combining Bayesian reasoning with causal inference.

  • GitHub: https://github.com/pymc-labs/CausalPy

  • WebPage: https://www.pymc-labs.io/blog-posts/causalpy-a-new-package-for-bayesian-causal-inference-for-quasi-experiments/

  • Blog:

    • https://www.pymc-labs.io/blog-posts/causalpy-a-new-package-for-bayesian-causal-inference-for-quasi-experiments/

    • https://towardsdatascience.com/causal-python-elon-musks-tweet-our-googling-habits-bayesian-synthetic-control-187114fc4aa8

  • Documentation: https://causalpy.readthedocs.io/en/latest/

  • Gallery: https://www.pymc.io/projects/examples/en/latest/gallery.html

• FLAML (Microsoft Research): for estimator and hyperparameter search

  • GitHub: https://microsoft.github.io/FLAML/

Reference:

• https://www.microsoft.com/en-us/research/group/causal-inference/

• https://www.microsoft.com/en-us/research/publication/tutorial-on-causal-inference-and-counterfactual-reasoning/

• https://github.com/microsoft/causica

• https://causalpython.io/