Fundamentals of Machine Learning: Unsupervised Learning

Active learning works for unsupervised learning because students need to physically experience how algorithms impose structure on unlabeled data. Moving bodies, plotting points, and comparing visuals help learners internalize that patterns emerge from mathematical choices, not objective truths.

• 12th Grade
• Computer Science
• 4 activities
• 15 min–40 min

Learning Objectives

1. Classify data points into distinct groups based on inherent similarities using clustering algorithms.
2. Compare the effectiveness of k-means and hierarchical clustering for different dataset structures.
3. Analyze the trade-offs between information loss and dimensionality reduction using techniques like PCA.
4. Evaluate the suitability of unsupervised learning methods for anomaly detection in financial transaction data.
5. Design a process to visualize high-dimensional data by applying dimensionality reduction techniques.

Problem-Based Learning

⏱ 25 min·Whole Class

Simulation Activity: Human K-Means Clustering

Tape a large coordinate grid on the floor. Give each student a card with (x, y) values and have them stand at their position. The teacher randomly assigns two students as initial centroids. Students assign themselves to the nearest centroid by walking toward it, then recompute centroids as a group average. Repeat for two more rounds. Students observe convergence and discuss whether the result is globally optimal.

Prepare & details

Explain how unsupervised learning can discover patterns without explicit labels.

Facilitation Tip: During the Human K-Means Clustering activity, have students physically walk to new centroid positions step-by-step rather than jumping to final clusters immediately.

Setup: Groups at tables with access to research materials

Materials: Problem scenario document, KWL chart or inquiry framework, Resource library, Solution presentation template

• Analyze
• Evaluate
• Create
• Decision-Making
• Self-Management
• Relationship Skills

Collaborative Problem-Solving

⏱ 40 min·Pairs

Collaborative Problem-Solving: Clustering Unlabeled Data

Students run k-means on a dataset of their choice, customer purchase data, penguin measurements, or movie ratings, using Python and scikit-learn. They experiment with different values of k, visualize the results, and write a paragraph interpreting what each cluster might represent. The ambiguity of interpreting unlabeled clusters is a key learning moment.

Prepare & details

Compare the applications of clustering and dimensionality reduction in data analysis.

Facilitation Tip: In the Clustering Unlabeled Data lab, model how to interpret silhouette scores by comparing two different k values side-by-side with the class.

Setup: Groups at tables with problem materials

Materials: Problem packet, Role cards (facilitator, recorder, timekeeper, reporter), Problem-solving protocol sheet, Solution evaluation rubric

• Apply
• Analyze
• Evaluate
• Create
• Relationship Skills
• Decision-Making
• Self-Management

Think-Pair-Share

⏱ 15 min·Pairs

Think-Pair-Share: Is This Clustering Useful?

Present two clustering results for the same dataset, one with two clusters, one with eight. Pairs discuss which is more useful for a specific business decision (e.g., designing a marketing campaign). There is no single right answer; the discussion surfaces the fact that 'good' clustering depends on the question being asked, not just on a mathematical metric.

Prepare & details

Analyze the challenges of evaluating the performance of unsupervised learning models.

Facilitation Tip: For the Gallery Walk of dimensionality reduction, assign each group one technique so they become the ‘experts’ who explain trade-offs to peers.

Setup: Standard classroom seating; students turn to a neighbor

Materials: Discussion prompt (projected or printed), Optional: recording sheet for pairs

• Understand
• Apply
• Analyze
• Self-Awareness
• Relationship Skills

Gallery Walk

⏱ 18 min·Small Groups

Gallery Walk: Dimensionality Reduction Visualization

Post printouts showing the same dataset in 3D and as a 2D PCA projection, alongside visualizations of t-SNE and UMAP. Students annotate each with what information appears preserved and what appears lost. The walk helps students understand dimensionality reduction as a compression decision with trade-offs rather than as a magical reveal of hidden truth.

Prepare & details

Explain how unsupervised learning can discover patterns without explicit labels.

Setup: Wall space or tables arranged around room perimeter

Materials: Large paper/poster boards, Markers, Sticky notes for feedback

• Understand
• Apply
• Analyze
• Create
• Relationship Skills
• Social Awareness

Teaching This Topic

Start with the Human K-Means activity to ground the concept in embodied learning. Then use the lab to connect abstract metrics like inertia to concrete decisions. Avoid rushing to applications before students have wrestled with how algorithms ‘see’ data differently than humans do. Research shows visual and kinesthetic experiences reduce misconceptions about objectivity in clustering results.

What to Expect

Successful learning looks like students articulating why different distance metrics or cluster counts produce different groupings. They should connect mathematical assumptions to real-world outcomes and critique when unsupervised methods are appropriate or misleading.

Watch Out for These Misconceptions

Common Misconception

During Human K-Means Clustering, watch for students assuming the final clusters reveal the ‘true’ groups in the data.

What to Teach Instead

Pause after two iterations and ask groups to compare their current centroids to the starting points, explicitly naming the mathematical assumption (e.g., Euclidean distance, fixed k) that shaped the shift.

Common Misconception

During the Clustering Unlabeled Data lab, watch for students believing clustering is only useful when labels are unknown.

What to Teach Instead

Ask students to overlay the ground-truth labels (if available) onto their clusters and calculate the adjusted Rand index to quantify overlap, then discuss why mismatch occurs even with clear patterns.

Common Misconception

During the Gallery Walk: Dimensionality Reduction Visualization, watch for students concluding that reduced dimensions always lose critical information.

What to Teach Instead

Have each group plot the same two principal components twice—once with original data and once with synthetic noise added—then ask them to identify which patterns persist, linking variance retention to noise reduction.

Assessment Ideas

Quick Check

After the Human K-Means Clustering activity, present students with a new unlabeled scatter plot. Ask them to sketch potential clusters, label their centroids, and explain the distance metric they implicitly used. Collect responses to identify students who default to Euclidean assumptions.

Discussion Prompt

After the Clustering Unlabeled Data lab, pose a scenario where students must choose between k-means and hierarchical clustering for a dataset of patient symptoms. Have them defend their choice using inertia scores and known label overlap, assessing their ability to connect metrics to real decisions.

Exit Ticket

During the Gallery Walk, give each student a sticky note to write one insight about how dimensionality reduction affects cluster separation. Review notes to see if students recognize that noise reduction can sharpen patterns, not just obscure them.

Extensions & Scaffolding

• Challenge early finishers to rerun the clustering lab using a density-based algorithm (DBSCAN) and compare results to k-means.
• Scaffolding for struggling students: provide a partially clustered scatter plot and ask them to adjust centroids manually before running code.
• Deeper exploration: have students apply PCA to the same dataset twice—once with all features and once with only the top two principal components—then visualize both to see how variance retention affects patterns.

Key Vocabulary

Clustering	An unsupervised learning technique that groups data points into clusters based on their similarity, without prior knowledge of group labels.
Centroid	The center of a cluster, typically calculated as the mean of all data points within that cluster, used in algorithms like k-means.
Dimensionality Reduction	A process that reduces the number of random variables under consideration by obtaining a set of principal variables, simplifying data while retaining essential information.
Principal Component Analysis (PCA)	A statistical procedure that uses an orthogonal transformation to convert a set of observations of possibly correlated variables into a set of values of linearly uncorrelated variables called principal components.
Unlabeled Data	Data that does not have predefined categories or tags, requiring algorithms to discover patterns or structures independently.