What does “data” look like? If you try to picture it, you probably see rows and columns - a spreadsheet or CSV, that can be conveniently loaded with pandas and cleaned and analyzed from there. As a data scientist, this will often be the form you want your data to be in - but it’s probably not how your data started.

Most modern data is generated automatically by human interaction with a web-backed application - every app they take, every click they make, all travels over a network and is saved by the server. Though in the rawest of forms this may be a log file, in most cases where it really goes is a database.

So, what is a database? A place for data! If it’s relational, it’s actually still pretty close to that rows and columns picture, though with some important additional functionality. These databases are commonly accessed using SQL - Structured Query Language - a standard based on relational algebra, and a useful tool known not just by data scientists but by software engineers, MBAs, and more.

If it’s so-called “NoSQL”, then it’s most likely a document-oriented database (or document store) - which, despite the glamor, is essentially a bunch of key-value pairs.