msuzen / looper Goto Github PK

A resource list for causality in statistics, data science and physics

statistics statistical-inference bayesian-inference causality-analysis causality data-science physics machine-learning causal-inference meta-learning

looper's Introduction

looper : A resource list for causality in statistics, data science and physics

Our honour to be mentioned by Judea Pearl on twitter.

 
      Aspiration to learn everything from data alone has 
      kept the ML community away from science.
      
      Judea Pearl

A resource list, code snippets or small scale software solutions for causal analysis in different areas. The name is inspired from the movie looper, which has a premise of time-like loops, probably the most complex causal subject from physics point of view.

The main entry is a markdown file as follows, any looper specific internal examples are lined there too :

A resource list for causality in statistics, data science and physics Other fields of course such as econometrics, epidemiology and many more.

Looper Nuggets

Looper Nuggets mimick a glossary of terms and concepts in causal inference, though they are entry to understanding concepts in pedagogical manner. See the list of them here.

License

This repository and all contributions are licensed under

Citing the repo

Please attribute this work as follows

@misc{suezen2018a,
  author = {Mehmet S{\"u}zen et. al.},
  title = {A resource list for causality in statistics, data science and physics},
  year = {2018},
  publisher = {GitHub},
  journal = {GitHub repository},
  howpublished = {\url{https://github.com/msuzen/looper}},
}

If you are embedding specific release, use the version link, for example for https://github.com/msuzen/looper/tree/v0.1.2 in howpublished tag.

Contributions

Please send a pull request or create an issue for suggestions or codes. See Basic how to contribute guide

looper's People

Contributors

Stargazers

Watchers

looper's Issues

add thesis from Arkansas

Achieving Causal Fairness in Machine Learning
Wu, Yongkai. University of Arkansas, ProQuest Dissertations Publishing, 2020. 27960037.

https://search.proquest.com/openview/5b94283bd4da8edc1b14bff3db4c9e77/1?pq-origsite=gscholar&cbl=18750&diss=y

Stefan Wager class notes

https://statweb.stanford.edu/~owen/courses/363/stats361.pdf
https://statweb.stanford.edu/~owen/courses/363/

fairness in Bayesian networks

https://deepmind.com/blog/article/Causal_Bayesian_Networks

broken link Causality: Models, Reasoning and Inference (2009) 2nd Edition.

Causality: Models, Reasoning and Inference (2009) 2nd Edition.
amzn.

shalizi notebook on discovery.

New MS tools

recent updates to DoWhy and EconML and introduce new tools DECI and ShowWhy. DECI (Deep End-to-end Causal Inference) combines causal discovery and inference; and ShowWhy provides a no-code interface to make causal inference easier for data analysts.
three marvelous packages from Microsoft – DoWhy, EconML, and FLAML

conditional average treatment effect (CATE)

CATE papers; main ones.

CV : https://usaito.github.io/files/ICML2020_CFCV.pdf

Bareinboim's RL works/section

New section on RL.

potential outcomes and ref therein.

Causal Inference Using Potential Outcomes: Design, Modeling, Decisions
Donald B Rubin
Journal of the American Statistical Association; Mar 2005; 100, 469; ABI/INFORM Global

causal entropy

https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.110.168702

A Python example for Gragner causality

explainers with causality methods

software: https://pair-code.github.io/what-if-tool/index.html

workshop: "Foundations and new horizons for causal inference"

https://www.ems-ph.org/JOURNALS/show_abstract.php?issn=1660-8933&vol=16&iss=2&rank=11

do-calculus links

add more

Baseless claim "... the fact that the PCM is not as well suited to analyzing structures that exhibit features of central interest to economists and econometricians.."
They omit the Elephant in the room that the first causal inference approach taken by Haavelmo - Economics Nobel

https://www.mitpressjournals.org/doi/full/10.1162/NECO_a_00295

The first two references translate Pearl's causal graphs to time-series data: https://advances.sciencemag.org/content/5/11/eaau4996
https://aip.scitation.org/doi/full/10.1063/1.5025050
This is an unofficial literature review of causality in time-series data: https://towardsdatascience.com/inferring-causality-in-time-series-data-b8b75fe52c46
You might find interesting that Pearl suggested the Sure-Thing Principle as an axiomatic basis for Bayesian inference: https://ftp.cs.ucla.edu/pub/stat_ser/r466.pdf
Finally, here's a quantum-like interpretation of the violations of the Sure-Thing principle: https://www.frontiersin.org/articles/10.3389/fpsyg.2016.00011/full