The Data Science Handbook, 2nd Edition

Preface to the Second Edition

In the first edition of this book, I called the introduction “Becoming a Unicorn.” Data science was a new field that was poorly understood, and data scientists were often called “unicorns” in reference to their miraculous ability to do both math and programming. I wrote the book with one central message: data science isn’t as inaccessible as people are making it out to be. It is perfectly reasonable for somebody to acquire the whole palette of skills required, and my book aspired to be a one‐stop‐shop for people to learn them.

A great deal has changed since then, and I’m delighted that the educational system has caught on. There are now degree programs and bootcamps that can teach the essentials of data science to most anybody who is willing to learn them. There are relatively standard curricula, fewer people who are baffled by the subject, and more young professionals embarking on this exciting career. Data science has gone from being an obscure priesthood to an exciting career that normal people can have.

As the discipline has expanded, the tools have also evolved, and I felt that a second edition was in order. By far the most important change I have made is more coverage of deep learning: previously I barely touched on RNNs, but now I continue up through topics such as encoder–decoder architectures, diffusion models, LLMs, and prompt engineering. AI tools are coming of age (perhaps AI is now where data science was 10 years ago) and a data scientist ...