Silhouette Score Calculator
Cluster cohesion vs separation for one point — the building block of the silhouette metric for choosing K.
Average the per-point silhouette over all points to score a whole clustering, and compare across K to pick the number of clusters. A negative silhouette means the point is closer to a neighboring cluster than its own — a sign of over- or under-clustering.
Formula
About Silhouette Score Calculator
The silhouette coefficient measures how well a single point fits its assigned cluster versus the nearest alternative, combining cohesion (how close it is to its own cluster) and separation (how far from the next-nearest) into one number from −1 to +1. This calculator computes it for one point from its two mean distances, building the intuition behind the full silhouette metric — averaged over all points, it's the standard way to evaluate a clustering and, by comparing across candidate K values, to choose the number of clusters when you have no ground truth labels.
How to use Silhouette Score Calculator
- 1Enter your values into Silhouette Score Calculator — sensible, domain-typical defaults are pre-filled so you see a real result immediately.
- 2The result recomputes live using the formula shown on the page; there is no button to press.
- 3Adjust any input to compare scenarios, then read the worked example to see the substituted numbers.
Why use Silhouette Score Calculator?
- ✓Computes Silhouette Score instantly in your browser — no sign-up, no upload, no server round-trip.
- ✓100% free and unlimited, with the exact formula shown: s = (b − a) / max(a, b) — a = mean intra-cluster distance, b = mean nearest-other-cluster distance; ranges −1 to +1.
- ✓Runs entirely client-side, so every value you enter stays private on your device.
- ✓Live recompute as you type, with a worked example and authoritative references for trust.
Frequently asked questions
How do I use silhouette scores to pick the number of clusters?+
Run clustering for several K values, compute the mean silhouette over all points for each, and pick the K with the highest average. A silhouette plot (per-cluster distributions) is even better — it reveals whether one cluster is dragging the average down or whether clusters are uneven in quality.
What does a negative silhouette mean?+
The point is, on average, closer to a neighboring cluster than to its own — it's probably misassigned. A few negatives are normal at boundaries; many negatives signal that your K is wrong or the clusters genuinely overlap and aren't well-separated in this feature space.
Silhouette vs elbow method for choosing K?+
The elbow method (inertia vs K) is heuristic and often ambiguous — the 'elbow' is subjective. Silhouette gives an actual quality score you can maximize, and it penalizes both under- and over-clustering. Many practitioners use both; silhouette is the more principled tiebreaker.
Does silhouette work for all clustering algorithms?+
It works for any algorithm that assigns points to clusters and any distance metric, but it implicitly favors convex, roughly equal-sized clusters — the kind K-means produces. For density-based clusters (DBSCAN) with arbitrary shapes, silhouette can be misleading; consider density-based validity indices instead.
Related tools
- Expected Calibration Error (ECE) Calculator
- Top-K Accuracy Calculator
- MAPE & sMAPE Calculator
- MASE (Mean Absolute Scaled Error) Calculator
- Normalized Mutual Information Calculator
- Recommender Hit Rate & Coverage Calculator
- Medical Diagnostic Test — Confusion Matrix & Metrics Calculator
- Fraud Detection — Confusion Matrix & Metrics Calculator
Related ML & AI tools
ROC-AUC Calculator (from TPR/FPR points)
Trapezoidal area under the ROC curve from your (FPR, TPR) operating points — the threshold-independent ranking score.
● LiveClassification Threshold Cost Calculator
Find the probability cutoff that minimizes expected cost given your false-positive and false-negative penalties.
● LiveAdjusted Rand Index Calculator
Agreement between two clusterings (or clustering vs ground truth), corrected for chance — from pair counts.
● Live