COMP 2306 · Data Science: Cluster Analysis

Led by Karl Pearson Simulacrum

K-means, hierarchical clustering, and market segmentation — finding natural groups in unlabelled data. Led by the inventor of the correlation coefficient.

1. Module 1

   What Is Clustering and Why Does It Matter?

   Led by Karl Pearson Simulacrum

   The question

   A cluster is a group of observations that are closer to each other than to members of other groups. Why must you standardise features before measuring distance — and when would you use clustering instead of classification?

   Outcome

   The student can explain clustering, explain why standardisation matters, and distinguish it from classification.

   Sub-units

   1. ○ 1.1 Clustering vs Classification Open this unit → ✓ ↻
   Begin the course → Print certificate of completion
2. Module 2

   K-Means Clustering

   Led by Karl Pearson Simulacrum

   The question

   K-means minimises within-cluster sum of squares — the total distance from each point to its cluster centre. The Elbow Method plots WCSS vs k. What does the elbow actually represent, and what happens when there is no elbow?

   Outcome

   The student can implement K-means, apply the Elbow Method, and visualise clusters.

   Sub-units

   1. ○ 2.1 Mall Segmentation Open this unit → ✓ ↻
   Open this module → Print certificate of completion
3. Module 3

   Hierarchical Clustering

   Led by Karl Pearson Simulacrum

   The question

   K-means requires k in advance. Hierarchical clustering produces a dendrogram from which you choose the level of granularity after seeing the data. How do you read a dendrogram — and when do you cut it?

   Outcome

   The student can produce and read a dendrogram and compare hierarchical and K-means results.

   Sub-units

   1. ○ 3.1 Dendrogram Analysis Open this unit → ✓ ↻
   Open this module → Print certificate of completion
4. Module 4

   Clustering for Market Segmentation

   Led by Karl Pearson Simulacrum

   The question

   Four customer segments: Fans (satisfied and loyal), Roamers (satisfied but disloyal), Supporters (loyal but dissatisfied), Alienated. What strategy does each require — and how do you validate that these segments are real?

   Outcome

   The student can execute market segmentation, interpret centroids as archetypes, and connect findings to strategy.

   Sub-units

   1. ○ 4.1 Market Segmentation Analysis Open this unit → ✓ ↻
   Open this module → Print certificate of completion
5. Module 5

   Cluster Evaluation and the Limits of Unsupervised Learning

   Led by Karl Pearson Simulacrum

   The question

   "How do I know these four groups really exist?" Silhouette score, stability, business validation. What do each of these tests actually establish — and when should you trust a clustering result?

   Outcome

   The student can evaluate clustering quality and explain what makes a clustering result trustworthy.

   Sub-units

   1. ○ 5.1 Final Essay: Are These Clusters Real? Open this unit → ✓ ↻
   Open this module → Print certificate of completion