Brian Godsey [Brian Godsey] - Think Like a Data Scientist: Tackle the data science process step-by-step

Data science still carries the aura of a new field. Most of its componentsstatistics, software development, evidence-based problem solving, and so ondescend directly from well-established, even old, fields, but data science seems to be a fresh assemblage of these pieces into something that is new, or at least feels new in the context of current public discourse.

Like many new fields, data science hasnt quite found its footing. The lines between it and other related fieldsas far as those lines matterare still blurry. Data science may rely on, but is not equivalent to, database architecture and administration, big data engineering, machine learning, or high-performance computing, to name a few.

The core of data science doesnt concern itself with specific database implementations or programming languages, even if these are indispensable to practitioners. The core is the interplay between data content, the goals of a given project, and the data-analytic methods used to achieve those goals. The data scientist, of course, must manage these using any software necessary, but which software and how to implement it are details that I like to imagine have been abstracted away, as if in some distant future reality.

This book attempts to foresee that future in which the most common, rote, mechanical tasks of data science are stripped away, and we are left with only the core: applying the scientific method to data sets in order to achieve a projects goals. This, the process of data science, involves software as a necessary set of tools, just as a traditional scientist might use test tubes, flasks, and a Bunsen burner. But, what matters is whats happening on the inside: whats happening to the data, what results we get, and why.

In the following pages, I introduce a wide range of software tools, but I keep my descriptions brief. More-comprehensive introductions can always be found elsewhere, and Im more eager to delve into what those tools can do for you, and how they can aid you in your research and development. Focus always returns to the key concepts and challenges that are unique to each project in data science, and the process of organizing and harnessing available resources and information to achieve the projects goals.

To get the most out of this book, you should be reasonably comfortable with elementary statisticsa college class or two is fineand have some basic knowledge of a programming language. If youre an expert in statistics, software development, or data science, you might find some parts of this book slow or trivial. Thats OK; skip or skim sections if you must. I dont hope to replace anyones knowledge and experience, but I do hope to supplement them by providing a conceptual framework for working through data science projects, and by sharing some of my own experiences in a constructive way.