book

Causal Inference for Data Science

Name: Causal Inference for Data Science
Author: Aleix Ruiz de Villa
ISBN: 9781633439658

by Aleix Ruiz de Villa

January 2025

Beginner to intermediate

392 pages

12h 19m

English

Manning Publications

Read now

Unlock full access

Causal Inference for Data Science
copyright
dedication
contents
preface
acknowledgments
about this book
PrerequisitesHow this book is organized: A road mapThe learning path and philosophy of this bookDifferent learning stylesDeveloping intuition and formal methodologyBuilding your intuitionPracticing the methodologyAbout the codeliveBook discussion forum
about the author
about the cover illustration
Part 1 Inference and the role of Confounders

1 Introducing causality
1.1 How causal inference works1.1.1 Step 1: Determine the type of data1.1.2 Step 2: Understand your problem1.1.3 Step 3: Create a model1.1.4 Step 4: Share your model1.1.5 Step 5: Apply causal inference techniques1.2 Contrasts between causal models and the predictive models of machine learning1.3 Experimental studies1.3.1 Motivating example: Deploying a new website1.3.2 A/B testing1.3.3 Randomized controlled trials1.3.4 Steps to perform an A/B test1.3.5 Limitations of A/B testing and RCTs1.4 Observational studies1.4.1 Simulating synthetic data1.4.2 Causal effects under confounding1.5 Reviewing basic statistical concepts1.5.1 Empirical distributions and data-generating distributions1.5.2 A refresher on conditional probabilities and expectations1.6 Further reading1.7 Chapter quizSummary
2 First steps: Working with confounders
2.1 Learning the basic elements of causal inference through Simpson’s paradox2.1.1 What’s the problem?2.1.2 Develop your intuition: How to approach the problem2.1.3 Solving Simpson’s paradox2.2 Generalizing to other problems2.2.1 Describing the problem with a graph2.2.2 Articulating what we would like to know2.2.3 Finding the way to calculate the causal effect2.2.4 Articulating what we would like to know: The language of interventions2.2.5 Finding the way to calculate the causal effect: The adjustment formula2.2.6 How does the treatment work in each situation? The positivity assumption2.3 Interventions and RCTs2.4 First contact with the structural approach2.4.1 Simulating the kidney stone example2.4.2 Interventions in the structural approach2.5 When to apply the adjustment formula2.5.1 RCT or A/B test2.5.2 Confounders2.5.3 Unobserved confounders2.5.4 Mediators2.5.5 Many confounders2.5.6 Outcome predictive variables2.5.7 Treatment predictive variables2.5.8 Conditional Intervention2.5.9 Combining all the previous situations2.5.10 Summarizing the differences between intervening and applying the adjustment formula2.6 So, what’s the plan?2.7 Lessons learned2.8 Chapter quizSummary
3 Applying causal inference
3.1 When and why to use graphs in causal inference analysis3.2 Steps to formulate your problem using graphs3.2.1 List all the variables3.2.2 Create your graph3.2.3 State your assumptions3.2.4 State your objectives3.2.5 Check the positivity assumption3.3 Other examples3.3.1 Recommender systems3.3.2 Pricing3.3.3 Simulations3.4 Further reading3.5 Chapter quizSummary
4 How machine learning and causal inference can help each other
4.1 What does supervised learning do?4.1.1 When should causal inference be used vs. supervised learning?4.1.2 The goal of data fitting4.1.3 When the future and the past behave the same way4.1.4 When do causal inference and supervised learning coincide?4.1.5 Predictive error is a false friend4.1.6 Validation of interventions4.2 How does supervised learning participate in causal inference?4.2.1 Empirical and generating distributions in the adjustment formula4.2.2 The flexibility of the adjustment formula4.2.3 The adjustment formula for continuous distributions4.2.4 Algorithms for calculating the adjustment formula4.2.5 Cross–fitting: Avoiding overfitting4.3 Other applications of causal inference in machine learning4.3.1 Reinforcement learning4.3.2 Fairness4.3.3 Spurious correlations4.3.4 Natural language processing4.3.5 Explainability4.4 Further reading4.5 Chapter quizSummary
Part 2 The adjustment formula in practice
5 Finding comparable cases with propensity scores
5.1 Developing your intuition about the propensity scores5.1.1 Finding matches for estimating causal effects5.1.2 But is there a match?5.1.3 Why matching can be hard5.1.4 How propensity scores can be used to calculate the ATE5.2 Basic notions of propensity scores5.2.1 Which cases are we working with?5.2.2 What are the propensity scores?5.2.3 The positivity assumption is … an assumption5.3 Propensity scores in practice5.3.1 Data preparation5.3.2 Calculating the propensity scores5.3.3 Assess the positivity assumption5.3.4 Calculating ATEs drawn from the propensity scores5.4 Calculating propensity score adjustment: An exercise5.4.1 Exercise steps5.5 Further reading5.6 Chapter quizSummary
6 Direct and indirect effects with linear models
6.1 Estimating causal effects with linear models6.1.1 Simulating a pricing problem: A walkthrough6.1.2 Direct and indirect effects6.2 Understanding causal dynamics through linear models6.2.1 The analogy of a gas flowing through pipes6.2.2 How correlation flows through a graph6.2.3 Calculating causation and correlation from the arrows' coefficients6.2.4 Linear models and the “do” operator6.3 Chapter quizSummary
7 Dealing with complex graphs
7.1 Altering the correlation between two variables conditioning on a third one7.1.1 Arrival time example of conditional independence7.1.2 Mathematical example of conditional independence7.1.3 Breaking a causal model into independent modules7.1.4 The bricks of DAGs: Factorizing probability distributions7.1.5 What’s the d-separation about?7.1.6 Defining d-separation7.2 Back-door criterion7.2.1 The importance of the back-door criterion7.3 Good and bad controls7.3.1 Good controls7.3.2 Neutral controls7.3.3 Bad controls7.4 Revisiting previous chapters7.4.1 Efficient controls7.4.2 Propensity score7.4.3 Again: Don’t include variables in your model just because they make the model more accurate7.4.4 Should you adjust for income?7.4.5 Linear models7.5 An advanced tool for identifying causal effects: The do-calculus7.6 Further reading7.7 Chapter quizSummary
8 Advanced tools with the DoubleML library
8.1 Double machine learning8.1.1 FWL theorem: The predecessor of DML8.1.2 Nonlinear models with DML8.1.3 DML in practice8.1.4 Heterogeneous treatment effects8.2 Confidence intervals8.2.1 Simulating new datasets with bootstrapping8.2.2 Analytical formulas for confidence intervals8.3 Doubly robust estimators8.3.1 AIPW in practice8.4 Further reading8.5 Chapter quizSummary
Part 3 Other strategies beyond the adjustment formula
9 Instrumental variables
9.1 Understanding IVs through an example9.1.1 The example’s DAG9.1.2 IV assumptions9.1.3 IVs in RCTs9.2 Estimating the causal effect with IVs9.2.1 Applying IVs with linear models9.2.2 Applying IVs for partially linear models9.2.3 An alternative formula for the IV method9.2.4 The lack of a general formula for the general IV graph9.3 Instrumental variables in practice9.3.1 Two-stage least squares (2SLS) algorithm9.3.2 Weak instruments9.3.3 IVs with DoubleML9.4 References9.5 Chapter quizSummary
10 Potential outcomes framework
10.1 What is a potential outcome?10.1.1 Individual outcomes10.1.2 Population outcomes10.1.3 Causal effects10.1.4 PO assumptions10.2 How do POs relate to DAGs?10.2.1 The first law of causal inference10.2.2 Expressing PO assumptions with DAGs10.2.3 Counterfactuals10.3 Adjustment formula with potential outcomes10.4 IVs with potential outcomes10.5 Chapter quizSummary
11 The effect of a time-related event
11.1 Which types of data will we use?11.2 Regression discontinuity design11.2.1 Data simulation11.2.2 RDD terminology11.2.3 Assumptions11.2.4 Effect estimation11.2.5 RDD in practice11.3 Synthetic controls11.3.1 Data simulation11.3.2 Synthetic controls terminology11.3.3 Assumptions11.3.4 Effect estimation11.3.5 Synthetic controls in practice11.3.6 Selecting training and predicting time periods11.4 Differences in differences11.4.1 Data simulation11.4.2 DiD terminology11.4.3 Assumptions11.4.4 Effect estimation11.4.5 In practice11.5 Chapter quiz11.6 Method comparison11.7 ReferencesSummary
appendix A The math behind the adjustment formula
appendix B Solutions to exercises in chapter 2
B.1 Solution to Simpson’s paradox for treatment BB.2 Observe and do are different thingsB.2.1 SolutionB.3 What do we need to adjust?B.3.1 RCTB.3.2 ConfounderB.3.3 Unobserved ConfounderB.3.4 MediatorsB.3.5 Outcome predictive variables
appendix C Technical lemma for the propensity scores
appendix D Proof for doubly robust estimator
D.1 DR property with respect to the T-learnerD.2 Doubly robust property with respect to inverse probability weighting
appendix E Technical lemma for the alternative instrumental variable estimator
appendix F Proof of the instrumental variable formula for imperfect compliance
index

Content preview from Causal Inference for Data Science

about the author

Aleix Ruiz de Villa Robert has over 15 years of experience as a data scientist. He earned his PhD in mathematics and a master’s in financial mathematics from Universitat Autònoma de Barcelona. He was the head of data science at LaVanguardia.com and SCRM (Lidl International Hub) and chief data scientist at Onna.

Since 2019, he has been working independently as a freelancer, focusing on research projects and teaching at various universities and business schools. He also advises the startup Nuclia.

Aleix is very active in organizing open events related to data science. He helped run the Barcelona R Users Group from 2011 to 2017 and founded and co-organized the Barcelona Data Science and Machine Learning meetup from 2014 ...

Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.

Read now

Unlock full access

More than 5,000 organizations count on O’Reilly

O’Reilly covers everything we've got, with content to help us build a world-class technology community, upgrade the capabilities and competencies of our teams, and improve overall team performance as well as their engagement.

Julian F.

Head of Cybersecurity

I wanted to learn C and C++, but it didn't click for me until I picked up an O'Reilly book. When I went on the O’Reilly platform, I was astonished to find all the books there, plus live events and sandboxes so you could play around with the technology.

Addison B.

Field Engineer

I’ve been on the O’Reilly platform for more than eight years. I use a couple of learning platforms, but I'm on O'Reilly more than anybody else. When you're there, you start learning. I'm never disappointed.

Amir M.

Data Platform Tech Lead

I'm always learning. So when I got on to O'Reilly, I was like a kid in a candy store. There are playlists. There are answers. There's on-demand training. It's worth its weight in gold, in terms of what it allows me to do.

Mark W.

Embedded Software Engineer

Publisher Resources

ISBN: 9781633439658Publisher Support Publisher Website Errata Page

Cloud Computing

Data Engineering

Data Science

AI & ML

Programming Languages

Software Architecture

IT/Ops

Security

Design

Business

Soft Skills

Causal Inference for Data Science

by Aleix Ruiz de Villa

about the author

Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.