Responsible Data Science

Explore the most serious prevalent ethical issues in data science with this insightful new resource
 
The increasing popularity of data science has resulted in numerous well-publicized cases of bias, injustice, and discrimination. The widespread deployment of "Black box" algorithms that are difficult or impossible to understand and explain, even for their developers, is a primary source of these unanticipated harms, making modern techniques and methods for manipulating large data sets seem sinister, even dangerous. When put in the hands of authoritarian governments, these algorithms have enabled suppression of political dissent and persecution of minorities. To prevent these harms, data scientists everywhere must come to understand how the algorithms that they build and deploy may harm certain groups or be unfair.
 
Responsible Data Science delivers a comprehensive, practical treatment of how to implement data science solutions in an even-handed and ethical manner that minimizes the risk of undue harm to vulnerable members of society. Both data science practitioners and managers of analytics teams will learn how to:
* Improve model transparency, even for black box models
* Diagnose bias and unfairness within models using multiple metrics
* Audit projects to ensure fairness and minimize the possibility of unintended harm
 
Perfect for data science practitioners, Responsible Data Science will also earn a spot on the bookshelves of technically inclined managers, software developers, and statisticians.

List of contents

Introduction xix
 
Part I Motivation for Ethical Data Science and Background Knowledge 1
 
Chapter 1 Responsible Data Science 3
 
The Optum Disaster 4
 
Jekyll and Hyde 5
 
Eugenics 7
 
Galton, Pearson, and Fisher 7
 
Ties between Eugenics and Statistics 7
 
Ethical Problems in Data Science Today 9
 
Predictive Models 10
 
From Explaining to Predicting 10
 
Predictive Modeling 11
 
Setting the Stage for Ethical Issues to Arise 12
 
Classic Statistical Models 12
 
Black-Box Methods 14
 
Important Concepts in Predictive Modeling 19
 
Feature Selection 19
 
Model-Centric vs. Data-Centric Models 20
 
Holdout Sample and Cross-Validation 20
 
Overfitting 21
 
Unsupervised Learning 22
 
The Ethical Challenge of Black Boxes 23
 
Two Opposing Forces 24
 
Pressure for More Powerful AI 24
 
Public Resistance and Anxiety 24
 
Summary 25
 
Chapter 2 Background: Modeling and the Black-Box Algorithm 27
 
Assessing Model Performance 27
 
Predicting Class Membership 28
 
The Rare Class Problem 28
 
Lift and Gains 28
 
Area Under the Curve 29
 
AUC vs. Lift (Gains) 31
 
Predicting Numeric Values 32
 
Goodness-of-Fit 32
 
Holdout Sets and Cross-Validation 33
 
Optimization and Loss Functions 34
 
Intrinsically Interpretable Models vs. Black-Box Models 35
 
Ethical Challenges with Interpretable Models 38
 
Black-Box Models 39
 
Ensembles 39
 
Nearest Neighbors 41
 
Clustering 41
 
Association Rules 42
 
Collaborative Filters 42
 
Artificial Neural Nets and Deep Neural Nets 43
 
Problems with Black-Box Predictive Models 45
 
Problems with Unsupervised Algorithms 47
 
Summary 48
 
Chapter 3 The Ways AI Goes Wrong, and the Legal Implications 49
 
AI and Intentional Consequences by Design 50
 
Deepfakes 50
 
Supporting State Surveillance and Suppression 51
 
Behavioral Manipulation 52
 
Automated Testing to Fine-Tune Targeting 53
 
AI and Unintended Consequences 55
 
Healthcare 56
 
Finance 57
 
Law Enforcement 58
 
Technology 60
 
The Legal and Regulatory Landscape around AI 61
 
Ignorance Is No Defense: AI in the Context of Existing Law and Policy 63
 
A Finger in the Dam: Data Rights, Data Privacy, and Consumer Protection Regulations 64
 
Trends in Emerging Law and Policy Related to AI 66
 
Summary 69
 
Part II The Ethical Data Science Process 71
 
Chapter 4 The Responsible Data Science Framework 73
 
Why We Keep Building Harmful AI 74
 
Misguided Need for Cutting-Edge Models 74
 
Excessive Focus on Predictive Performance 74
 
Ease of Access and the Curse of Simplicity 76
 
The Common Cause 76
 
The Face Thieves 78
 
An Anatomy of Modeling Harms 79
 
The World: Context Matters for Modeling 80
 
The Data: Representation Is Everything 83
 
The Model: Garbage In, Danger Out 85
 
Model Interpretability: Human Understanding for Superhuman Models 86
 
Efforts Toward a More Responsible Data Science 89
 
Principles Are the Focus 90
 
Nonmaleficence 90
 
Fairness 90
 
Transparency 91
 
Accountability 91
 
Privacy 92
 
Bridging the Gap Between Principles and Practice with the Responsible Data Science (RDS) Framework 92
 
Justific

About the author

GRANT FLEMING is a Data Scientist at Elder Research Inc. His professional focus is on machine learning for social science applications, model interpretability, civic technology, and building software tools for reproducible data science.
PETER BRUCE is the Senior Learning Officer at Elder Research, Inc., author of several best-selling texts on data science, and Founder of the Institute for Statistics Education at Statistics.com, an Elder Research Company.

Summary

Explore the most serious prevalent ethical issues in data science with this insightful new resource

The increasing popularity of data science has resulted in numerous well-publicized cases of bias, injustice, and discrimination. The widespread deployment of "Black box" algorithms that are difficult or impossible to understand and explain, even for their developers, is a primary source of these unanticipated harms, making modern techniques and methods for manipulating large data sets seem sinister, even dangerous. When put in the hands of authoritarian governments, these algorithms have enabled suppression of political dissent and persecution of minorities. To prevent these harms, data scientists everywhere must come to understand how the algorithms that they build and deploy may harm certain groups or be unfair.

Responsible Data Science delivers a comprehensive, practical treatment of how to implement data science solutions in an even-handed and ethical manner that minimizes the risk of undue harm to vulnerable members of society. Both data science practitioners and managers of analytics teams will learn how to:
* Improve model transparency, even for black box models
* Diagnose bias and unfairness within models using multiple metrics
* Audit projects to ensure fairness and minimize the possibility of unintended harm

Perfect for data science practitioners, Responsible Data Science will also earn a spot on the bookshelves of technically inclined managers, software developers, and statisticians.

Product details

• Authors: Peter C Bruce, Peter C. Bruce, Peter C. (Massachusetts Institute of Techno Bruce, Bruce Peter C., G Fleming, Gran Fleming, Grant Fleming, Grant (Massachusetts Institute of Technol Fleming, Grant Bruce Fleming, Fleming Grant
• Publisher: Wiley, John and Sons Ltd
• Languages: English
• Product format: Paperback / Softback
• Released: 30.06.2021
• EAN: 9781119741756
• ISBN: 978-1-119-74175-6
• No. of pages: 304
• Subjects: Natural sciences, medicine, IT, technology > IT, data processing > Application software; Informatik, Data Science, computer science, Data Mining & Knowledge Discovery, Data Mining u. Knowledge Discovery