Skip to content
Open Fraud LabsAcademy

Capstone project

Data Science from Scratch capstone

Bring the whole course together in one small, real analysis. Once it’s approved and all lesson quizzes are passed, you can claim your certificate.

The brief

Pick a public dataset that interests you (for example from Kaggle, the World Bank, or a government open-data portal). Then:

  1. Ask one clear question the data can help answer.
  2. Describe the data: source, rows and columns, data types, and the target if you build a model.
  3. Clean it: handle missing values, duplicates and obvious errors, and explain each choice.
  4. Explore it: summary statistics and at least three well-labelled charts.
  5. Model it (optional but encouraged): a simple model with a train/test split and a fair evaluation metric.
  6. Communicate it: state your answer, your evidence, and the limitations.

What to submit

  • A public GitHub repository with your notebook or code and a README that walks through the steps above.
  • A short write-up (at least 50 characters) in the form below: your question and what you found.

How it’s reviewed

Your capstone is peer reviewed: three other learners score it with a published rubric, and you review three capstones in return on the peer review page. The middle score counts and the pass mark is 70%. Open Fraud Labs can review any project. If changes are requested, improve it and resubmit as many times as you need.