Advanced level · Module 5: Data Governance and Responsible Data Science · Lesson 3 of 4
Reproducibility and Review
Make analyses reproducible with version control and managed environments, and review analysis code.
In this lesson you will
- Use Git and environment files for reproducible analysis
- Review analysis code for errors and clarity
Key concepts
Git for analysisenvironment filesrandom seedsanalysis code review
The lesson
Track code, notebooks and documentation with Git, and record package versions in an environment file so results can be reproduced later.
Fix random seeds, keep raw data read-only and generate every output from code.
Review analysis like software: a colleague checks joins, filters, weights and statistical choices. Many published errors come from a single wrong filter or merge.
Step by step
- Put the project under Git.
- Create an environment file.
- Rebuild outputs on a fresh environment.
- Review a colleague's notebook with a checklist.
Practice activity
Make the provided project reproducible and review a colleague's analysis.
Expected result: A reproducible project and a review with specific findings.
Check your understanding
-
Why record package versions?
Show answer
Results can change between versions; recording them allows exact reproduction.
-
What should an analysis review check?
Show answer
For example joins, filters, weights and statistical choices.
Taught live by an instructor — online or in person. Free course lessons are for self-study and do not lead to a certificate — see registration options.