Segmentation dashboard
Create unlabeled customers, discover natural groups, and explain their purchasing patterns.
Score0
Silhouette—
Davies–Bouldin—
Within-cluster SSE—
Outliers0%
Iteration0
Game setup and dataset controls
Feature selection and importance
Choose variables and adjust their influence before clustering.
Clustering and preprocessing options
Interactive clustering canvas
Click a customer to inspect it. In manual mode, click repeatedly to change its segment.
Drag to reposition • wheel to zoom
Generate a dataset, select features, and start clustering.
0
Business usefulness
and model quality
and model quality
Active challenge
Cluster Count Challenge
Find a useful number of customer groups
Balance separation, stability, and business usefulness without creating tiny clusters.
0 / 100
challenge points
Run a clustering model to receive feedback.
Selected customer and segment profiles
Inspect individual behavior and assign business-friendly segment names.
Select a customer on the canvas or table.
Plotly.js visual analysis
Compare separation, segment sizes, feature profiles, and cluster-count choices.
Customer data and segment assignments
Sort, inspect, and export the fictional dataset.
Learning guide
Unlabeled discovery
Clustering finds structure without known class labels. You interpret the groups after the algorithm finishes.
Why scaling matters
Large numerical ranges can dominate distances. Scaling gives selected features a more comparable influence.
Avoid false certainty
Clusters are exploratory patterns, not permanent identities. Validate stability and business meaning before acting.
Common mistakes
Too many clusters, ignored outliers, irrelevant features, and unscaled inputs can create misleading customer stories.