📌 So you’ve got a dataset, here’s how you clean it!

data cleaning
data practices
workflow
reusability
A practical, language-agnostic overview of data cleaning steps, useful not for its tooling, but for the clarity of thinking behind it.
Published

August 18, 2025

Came across this article again while skimming through my old bookmarks. I remembered liking it the first time around for how grounded and practical it was — and it still holds up.

📌 So You’ve Got a Dataset. Here’s How You Clean It.

A step-by-step mindset for approaching data cleaning, by Ang Li-Lian.

Reflections

What I appreciate about this piece is that it’s not tied to a specific tool or language. You could apply the exact same process in R, Python, SQL, or Excel, the core ideas are all about intentional, structured thinking.

Some solid reminders:

  • Start by actually looking at your dataset. Dimensions, types, nulls, duplicates, don’t skip the orientation.
  • Don’t treat missing data like an afterthought. Handle it with intention based on context.
  • Encode categorical variables early and deliberately. It’ll save you pain later.
  • Check your assumptions with basic summaries, sometimes the weirdness is obvious if you pause to look.
  • Modularize your cleaning steps. If you build them clearly, you can reuse them across projects.

It’s not fancy, but that’s the point. There’s something satisfying about a clean, thoughtful approach that isn’t trying to be clever, just effective. I’ve found this kind of steady rhythm in cleaning always pays off, especially when I come back to a dataset weeks or months later and can still follow my own trail.


This post builds on a recent LinkedIn #BookmarkDive reflection, feel free to join the conversation there.

Back to top