Interview: 'you're asked to segment 10M customers — how?'

hard

Answer

$(1) Sample down to 1M for prototype.$
$(2) Feature engineering: RFM (recency, frequency, monetary), behavioral counts, categorical embeddings.$
$(3) Scale (log-transform skewed features, z-score).$
$(4) PCA/UMAP to 20-50 dim for compute.$
$(5) Mini-batch k-means or HDBSCAN across candidate k (silhouette + business interpretability).$
$(6) Profile clusters with per-feature means + decision-tree surrogate.$
$(7) Stability check via bootstrap.$
$(8) Deliver actionable segments with names + business hypotheses.$

Check yourself — multiple choice

Just run k-means
Sample / feature-engineer (RFM) / scale / PCA-UMAP / mini-batch k-means or HDBSCAN + silhouette / profile with means + tree / stability bootstrap / name segments with hypotheses
Random
Only PCA

10M customer segmentation: sample + RFM + scale + reduce + cluster + profile + stability.

#clustering#interview#applications

214 interview questions in this topic.