Activity: Explore Structure Without Inventing Labels

Activity ID: ML-M04-A01 Estimated active time: 45-75 minutes

Goal

Produce the main Module 04 evidence: Exploration notebook with cluster/projection outputs and a written interpretation that includes at least three limits.

Inputs

• data/module_04_unsupervised_exploration_v1.csv
• data/data-dictionary.md
• notebooks/module-04-worked.ipynb
• notebooks/module-04-practice.ipynb

Instructions

1. Scale the features before clustering, and say why unscaled clustering would be dominated by one column.
2. Fit clusters at two or three different values of k.
3. For each k, describe what the clusters appear to separate — in plain English, not cluster numbers.
4. Check whether cluster membership tracks a column you already had. If it does, the clustering added nothing.
5. State what you would need to confirm these groups are real rather than an artefact of the algorithm.
6. Write two limitations specific to unsupervised results.

Quality checklist

• [ ] Features are scaled before distance is computed.
• [ ] More than one value of k was tried.
• [ ] Each cluster has a plain-English description.
• [ ] The note says whether clusters merely reproduce an existing column.
• [ ] Limitations acknowledge there is no ground truth to check against.

Submission format

Submit the notebook output or worksheet table plus a 4-6 sentence explanation.

Safety note

Use only supplied fictional or synthetic data. Do not add or upload real learner data.