K-Means Clustering

Clustering finds natural groups in unlabeled data. Slide k to change how many true clusters exist, then reveal them.

We have unlabeled data and want to find k=3 clusters. The algorithm needs a way to group these points automatically.

Each point is assigned to the nearest centroid — creating Voronoi-like regions. Press Play or drag the slider.

Centroids move to the mean of their assigned points — then we reassign. Watch the dashed arrows show displacement.

K-means converges when no points change their assignment — typically in 5-15 iterations. Watch inertia drop.

The elbow! At k=3, adding more clusters gives diminishing returns. This is usually the best choice.