Data Science Ethics Concepts, Techniques, and Cautionary Tales

Data science ethics is all about what is right and wrong when conducting data science. Data science has so far been primarily used for positive outcomes for businesses and society. However, just as with any technology, data science has also come with some negative consequences: an increase of privacy invasion, data-driven discrimination against sensitive groups, and decision making by complex models without explanations.

While data scientists and business managers are not inherently unethical, they are not trained to weigh the ethical considerations that come from their work - Data Science Ethics addresses this increasingly significant gap and highlights different concepts and techniques that aid understanding, ranging from k-anonymity and differential privacy to homomorphic encryption and zero-knowledge proofs to address privacy concerns, techniques to remove discrimination against sensitive groups, and various explainable AI techniques.

Real-life cautionary tales further illustrate the importance and potential impact of data science ethics, including tales of racist bots, search censoring, government backdoors, and face recognition. The book is punctuated with structured exercises that provide hypothetical scenarios and ethical dilemmas for reflection that teach readers how to balance the ethical concerns and the utility of data.

David Martens, Professor of Data Science, University of Antwerp, Belgium

David Martens is Professor of Data Science at the Department of Engineering Management, University of Antwerp, Belgium, and is also co-founder of the University of Antwerp spin-off Predicube. He teaches data mining and data science and ethics to postgraduate students studying business economics and
business engineering.

Foreword, Foster Provost
Preface
1. Introduction to Data Science Ethics
2. Ethical Data Gathering
3. Ethical Data Preprocessing
4. Ethical Modelling
5. Ethical Evaluation
6. Ethical Deployment
7. Conclusion