All projects
data
Clean a genuinely messy dataset
Take real-world data with all its problems and make it usable, documenting every decision.
Why it matters
Every data role is mostly cleaning. Showing that you document why you dropped a row is worth more than a chart built on data you did not question.
What you build
- Missing values handled deliberately
- Duplicates found and resolved
- Types and formats normalised
- A written record of every decision
If you want more
- Add automated quality checks
- Publish a cleaned dataset
Build it
Row counts, null rates, distinct values. Look before you clean.
- Fix types, dates and inconsistent categories
- Write down every decision and its reason
Material for the whole build
Shows you treat data decisions as decisions, not as defaults, which is what makes analysis trustworthy.
Get a plan built around projects like this