Michelle Blomberg

How Many Segments, Really?

Case study · AI Tools & Strategy · Learning Design

Marketing wants a fixed number of customer segments, and a colleague insists it is five. Your job is to defend the count the evidence actually supports, not the one that is convenient.

Module 6 · Clustering and PCA · summative simulation

A scatter plot of data points forming three colored clusters with a few ambiguous grey points between them

Run the simulation

Randomized on every run here, and deterministically seeded per learner when opened with a roster identifier, so an instructor can reproduce a graded attempt exactly. It runs in your browser and stores no personal data. Start in the panel below, or open it full screen →.

What this simulation is

This is the summative simulation for the clustering and PCA module. The learner is handed a customer dataset with genuine underlying structure, correlated noise columns, and non-spherical, unequally sized groups that punish naive k-means. The work is judged the way segmentation is really defended in industry: on cluster stability under bootstrap resampling, on agreement with the instructor’s private ground-truth labels the student never sees, and on the rigor of the cluster-count argument.

A confident answer with a weak justification scores worse than a cautious answer with strong evidence.

What it assesses

Why it resists AI substitution

The instructor’s private ground-truth labels and the injected structure of this dataset exist nowhere online, so a pasted generic answer cannot match them. Because the count must be defended under bootstrap stability tied to this exact data, a plausible-sounding but unverified number scores poorly.