The Data Science Handbook

Preface

This book was written to solve a problem. The people who I interview for data science jobs have sterling mathematical pedigrees, but most of them are unable to write a simple script that computes Fibonacci numbers (in case you aren't familiar with Fibonacci numbers, this takes about five lines of code). On the other side, employers tend to view data scientists as either mysterious wizards or used-car salesmen (and when data scientists can't be trusted to write a basic script, the latter impression has some merit!). These problems reflect a fundamental misunderstanding, by all parties, of what data science is (and isn't) and what skills its practitioners need.

When I first got into data science, I was part of that problem. Years of doing academic physics had trained me to solve problems in a way that was long on abstract theory but short on common sense or flexibility. Mercifully, I also knew how to code (thanks, Google™ internships!), and this let me limp along while I picked up the skills and mindsets that actually mattered.

Since leaving academia, I have done data science consulting for companies of every stripe. This includes web traffic analysis for tiny start-ups, manufacturing optimizations for Fortune 100 giants, and everything in between. The problems to solve are always unique, but the skills required to solve them are strikingly universal. They are an eclectic mix of computer programming, mathematics, and business savvy. They are rarely found together in one ...