K-Means clustering (k=3) repeatedly assigns every point to its nearest centroid and then relocates each centroid to the mean of its assigned points, thereby decreasing the objective ∑‖x − μ‖² at every iteration.