Home▸Data Science▸k-Means Lloyd Iterations: Monotone Inertia and Local Minima from Random Starts

k-Means Lloyd Iterations: Monotone Inertia and Local Minima from Random Starts

Step through Lloyd k-means on overlapping blobs: points take the colour of the nearest centroid, centroids move to the mean, and the inertia can only go down. Compare random and k-means++ seeding and see how different starts reach different local minima. A Blender-built data board with brute-force and conservation checks.

Data Science3DModerate60 FPS📱 Mobile-adapted🟠 Blender⇄ 2D version
k-Means Lloyd Iterations: Monotone Inertia and Local Minima from Random Starts ↗ Open standalone

Step through Lloyd k-means on overlapping blobs: points take the colour of the nearest centroid, centroids move to the mean, and the inertia can only go down. Compare random and k-means++ seeding and see how different starts reach different local minima. A Blender-built data board with brute-force and conservation checks.

⚙ Under the hood

Step through Lloyd k-means on overlapping blobs: points take the colour of the nearest centroid, centroids move to the mean, and the inertia can only go down. Compare random and k-means++ seeding and see how different starts reach different local minima. A Blender-built data board with brute-force and conservation checks.

K-MeansClusteringLloyd AlgorithmLocal MinimaInertia

3D · Three.js / WebGL renderer · 60 FPS target · runs fully client-side, no install

What did you find?

Add reproduction steps (optional)