K-Means clustering explorer

K-MEANS CLUSTERING EXPLORER

Step through assignment and centroid movement. Drag a centroid in 2D to see how initialization changes the same clustering in the full 3D cloud.

ready to assign

3D full feature cloud · drag to orbit

2D X × Y projection · drag centroids

Dataset Clusters K

Phaseassignment

Completed iterations0

Within-cluster error—

Cluster sizes—

Initialization matters: move a diamond before the first assignment, or reset repeatedly, and K-Means can settle on a different partition.

centroidunassigned pointassigned cluster