References

[Azariagmt]

Notebooks & Github codes

• Quickstart Notebook for using Causalgraphicalmodels python module: used to describe and manipulate Causal Graphical Models and Structural Causal Models.
• Introduction to CasualgraphicModel

Key Papers & blogs

• If correlation doesn’t imply causation, then what does?
• DoWhy: An End-to-End Library for Causal Inference (arxiv) - DoWhy package paper
• Mini course on Causality, Cambridge MIT
• Controlling Confounding Bias
• Slides on Causality
• Causality lecture note
• Confounding Bias
• Analysis of Breast Cancer Detection Using Different Machine Learning Techniques | SpringerLink
• ANALYSIS OF FEATURE SELECTION WITH CLASSFICATION: BREAST CANCER DATASETS

Talks & Videos

• Tutorial Session B – Causes and Counterfactuals: Concepts, Principles and Tools (Microft, 2014)
• Plenary 2: The Mathematics of Causal Inference: with Reflections on Machine Learning

General

• Causality by Judea Pearls 2nd edition
• Causal inference in statistics: An overview (2019)
• Main reference

Wikipedia

• Causal Graphical Models
• Structural Causal Models
• ​​Rubin causal model - Wikipedia
• Instrumental variables estimation - Wikipedia

[Azariagmt]

Discussion on wikipedia links

A good definition of causal graphs:
The causal graph can be drawn in the following way. Each variable in the model has a corresponding vertex or node and an arrow is drawn from a variable X to a variable Y whenever Y is judged to respond to changes in X when all other variables are being held constant. Variables connected to Y through direct arrows are called parents of Y, or "direct causes of Y," and are denoted by Pa(Y).

[Azariagmt]

Discussion on key papers and blogs

• correlation doesn’t imply causation.
• Simpson’s paradox is the fact that the correlation between two variables can actually be reversed when additional factors are considered. So two variables which appear correlated can become anticorrelated when another factor is taken into account. example of this is the civil rights act voting example and the kidney stone example
• Causal model image
  
• Causal model with some experimenter
  
• Pertrubed causal model: A causal model where the above experimenter model function is replaced with no new vertex
  

DoWhy
Python library that is built with causal assumptions as its firstclass citizens, based on the formal framework of causal graphs to specify and test causal assumptions. DoWhy presents an API for the four steps common to any causal analysis—

1. modeling the data using a causal graph and structural assumptions,
2. identifying whether the desired effect is estimable under the causal model
3. estimating the effect using statistical estimators, and finally
4. refuting the obtained estimate through robustness checks and sensitivity analyses.