Data science confusions and clarifications

Big data. Data science? Strategic data!

In October DevPro focuses on data analytics. The topic certainly deserves attention: data scientists capture top salaries, organizations spend well over $100 billion annually on big data globally, and big data is growing at three times the rate of the broader technology market. Whatever's happening, it's important enough to throw a lot of money at it.

As well-developed as data science appears to be, though, conversations about it still are often muddy. Rene Fassbender, for instance, sees the field moving in the direction of 'Smart Data' "... that take into account all relevant available information." Plenty of other commentators also emphasize goals of "understanding" or "insight" for data science.

As appealing as this vision is, I think it's misleading. Most organizations I see put a higher premium on action than understanding. "Data engineering" might be a better label for this use: the emphasis is on a practice that promotes profit. In this perspective, insight is at best an implementation detail. "Data engineering" has the potential for robust business results even in the "model-free' absence of strong theoretical understanding. It's not the only one, though; in some circles, data engineering is clearly subordinated to data science, the former a means to the end of the latter.

While a few-paragraph comment has no chance of reforming a billion-dollar industry, this article can sensitize you to how the people around you speak. When your department or company turns toward "data science", will success look more like understanding or profit? Do all the participants share that meaning? Is it better as a first step to improve one specific low-level practice, or does the organization need an executive-level strategic gain? Whatever fits those goals best, clarity about them can only help.