What Is Data Science?

Data science is the practice of turning raw, often messy data into answers someone can actually use — a mix of programming, statistics, and enough understanding of the subject matter to know which questions are worth asking in the first place.

Three skills, not one

Most definitions of data science describe it as sitting at the intersection of three things: programming (to actually get at and manipulate the data), statistics (to reason about it correctly instead of just eyeballing it), and domain knowledge (to know what a "good" or "surprising" result even looks like for the problem at hand). A brilliant statistician who doesn't understand the business is just as stuck as a great programmer who doesn't understand statistics — you genuinely need pieces of all three.

How it differs from plain programming

A typical programming task has a fairly clear spec: build a feature, fix a bug, the "correct" output is usually well-defined. Data science work usually starts from a fuzzier place — "why did signups drop last month?" — where the first job isn't writing code at all, it's figuring out what question to even ask, and what "answered" would look like.

How it differs from plain statistics

Classical statistics was largely built around small, carefully collected datasets and rigorous experimental design. Data science usually works with whatever data already exists — server logs, spreadsheets exported from three different systems, years of inconsistent records — and spends real effort just getting it into a shape that can be analyzed at all, before any statistical technique gets applied.

The unglamorous truth: most working data scientists will tell you that 70-80% of a real project is spent finding, cleaning, and understanding data — not building models. The modeling step people usually picture when they hear "data science" is often the smallest part of the actual work. The next lesson walks through the full workflow this really follows.
PreviousCourse Home