How do you interpret / visualize clusters?

medium

Answer

$(1) 2D projection with UMAP / t-SNE colored by cluster.$
$(2) Feature-importance-per-cluster: mean of each feature per cluster vs global mean \to describes each group.$
$(3) Decision-tree surrogate: fit a small tree predicting cluster label \to rule-based description.$
$(4) Representative sample per cluster (medoid + top / bottom feature scores).$
$Standard deliverable for a customer segmentation report.$

Check yourself — multiple choice

Just show numbers
UMAP/t-SNE projection + per-cluster feature means + decision-tree surrogate + medoid examples — standard segmentation reporting
Random
Only silhouette

Interpret clusters: projection + per-cluster feature stats + tree surrogate + medoids.

#clustering#visualization

214 interview questions in this topic.