Measure clustering compactness with fast centroid-based error analysis. Import grouped vectors, compare cluster spread, inspect point losses, and export polished summaries easily today.
| Point ID | Cluster | X | Y |
|---|---|---|---|
| P1 | A | 2 | 3 |
| P2 | A | 3 | 4 |
| P3 | A | 4 | 5 |
| P4 | B | 8 | 7 |
| P5 | B | 9 | 6 |
| P6 | B | 10 | 8 |
| P7 | C | 5 | 10 |
| P8 | C | 6 | 9 |
| P9 | C | 7 | 11 |
Within-cluster SSE measures how tightly points sit around their assigned cluster centroid. Lower values generally indicate more compact clustering.
For a point with features (x₁, x₂, ..., xₙ) and centroid (c₁, c₂, ..., cₙ):
The calculator sums point errors inside each cluster, then totals all cluster errors into one overall SSE value. It also reports mean squared error when normalization is enabled.
It measures how far cluster members are from their own centroid. A lower SSE usually means tighter and more compact clusters.
Lower SSE is usually better for compactness, but extremely low SSE can happen when too many clusters are used. Compare it with interpretability and model goals.
Yes. Add as many numeric feature columns as needed after the cluster column. The calculator handles multidimensional points automatically.
The calculator uses your centroid coordinates directly instead of computing cluster means. This is useful for validating existing clustering outputs.
Widely scattered points, outliers, poor cluster assignments, or badly placed centroids increase squared distances and push SSE upward.
Yes. It is ideal for checking k-means compactness, validating centroid quality, and comparing clustering runs across different k values.
Squaring emphasizes larger errors, keeps distances positive, and matches the standard objective minimized by k-means clustering.
Inspect that cluster for outliers, mixed patterns, scaling issues, or a missing feature transformation. It may need splitting or reassignment.
CSV format
Use a header row. The first two columns should be point id and cluster label.
Feature columns
Every remaining column must contain numeric values only.
Manual centroid format
Cluster label first, then centroid coordinates using the same feature order.
Best use case
Compare clustering compactness, review outliers, and document model quality.
Important Note: All the Calculators listed in this site are for educational purpose only and we do not guarentee the accuracy of results. Please do consult with other sources as well.