As a language, Python tends to be very forgiving of "bad code"; in fact, one of Python's core strenghts is that one can hack together something pretty terrible, and it'll mostly work and contribute to getting "real work" done. That said, when a codebase must be maintained and extended over a long time-horizon, maintainability concerns demand a better approach.

This tutorial is a live-coding exercise, covering common patterns seen when reviewing and refactoring code written by non-professional programmers, such as data scientists. It frames these as "disfluencies": code where the meaning is otherwise clear, where some goal is otherwise accomplished, but where the approach taken may be clumsy, prone to misunderstanding, or otherwise difficult to rely upon.