Running Clustering Algorithms

Use this skill when the main question is how to group unlabeled data points.

Overview

This skill covers algorithm choice, preprocessing implications, cluster validation, and interpretation for unsupervised segmentation problems.

When to Use This Skill

• Customer segmentation, cohort discovery, or grouping unlabeled records
• Choosing between centroid, density, or hierarchical clustering
• Reviewing silhouette score, Davies-Bouldin, or cluster stability

Not For / Boundaries

• Supervised prediction with labels: use training-machine-learning-models
• Pure anomaly review without clustering as the central method: use anomaly-detector
• Final narrative report packaging: use scientific-reporting

Typical Outputs

• Algorithm recommendation with parameter guidance