Silhouette Score Helper

Paste 2D or 3D points with cluster labels, get per-point silhouette scores and the overall mean to evaluate your clustering.

Data Points

One point per line: x,y,label or x,y,z,label

Results

Enter data points and click Compute

Score Reference

0.71 – 1.00

Strong structure — clusters are tight and well-separated.

0.26 – 0.70

Reasonable structure — some overlap or density variation.

Below 0.26

Weak or no structure — consider different k or algorithm.

Summary

Paste 2D or 3D points with cluster labels, get per-point silhouette scores and the overall mean to evaluate your clustering.

How it works

  1. Enter one data point per line in the format: x,y,label or x,y,z,label.
  2. Click "Compute" to calculate pairwise Euclidean distances between all points.
  3. For each point i, a(i) = mean distance to all other points in the same cluster.
  4. For each point i, b(i) = mean distance to all points in the nearest other cluster.
  5. Silhouette score s(i) = (b(i) - a(i)) / max(a(i), b(i)), ranging from -1 to +1.

Use cases

  • Evaluate whether k-means or DBSCAN produced tight, well-separated clusters.
  • Compare cluster counts by checking which k gives the highest mean silhouette score.
  • Identify misclassified points with negative silhouette scores for re-labeling.
  • Validate clustering results from a CSV export before reporting to stakeholders.

Frequently Asked Questions

Last updated: 2026-07-22 · Reviewed by Nham Vu