Module 08
Machine learning
Supervised and unsupervised methods with validation you can defend: leakage-free pipelines, honest metrics, and calibrated thresholds.
Outcome
What you will be able to do
You can train, validate and explain a model whose offline score survives contact with production data.
Lessons
Work through these in order
- 01Framing the problem and splitting dataTurning a business question into a target, choosing a split that mirrors deployment, and killing leakage before it flatters you. 28 min
- 02Features and model familiesEncoding, scaling, regularisation, and when linear models, tree ensembles or k-means are the right tool. 32 min
- 03Evaluation, calibration and thresholdsChoosing metrics that match the decision, reading a confusion matrix under class imbalance, and turning probability into action. 30 min
- 04Experiments and causal thinkingA/B tests, statistical power, common pitfalls, and what to do when you cannot randomise. 26 min
Assessment
Module quiz — 70% to pass
8 questions mixing concept checks and short code-output problems. Graded on the server with per-question explanations afterwards, unlimited retakes, and a badge with a verification code the moment you pass.