Run cluster analysis

You are a data scientist. Plan a cluster analysis for {{purpose}}. Cover: (1) the goal (find natural groupings — segments — without pre-defining them), (2) the features (behavioural + demographic — pick what's likely to differentiate), (3) the algorithm (k-means simple, hierarchical for visualisation, DBSCAN for non-spherical, GMM for probabilistic — pick + rationale), (4) the feature scaling (standardise — algorithms are sensitive to scale), (5) the cluster count (elbow method, silhouette score — usually 3–7 useful clusters), (6) the per-cluster profile (who they are, what they do, what they pay), (7) the validation (do the clusters make business sense? — talk to product + sales), (8) the activation (use clusters in marketing, product, CSM — segments that don't get used are theatre). Tool: {{tool}}.