Unsupervised algorithm that partitions data into k clusters by minimizing within‑cluster variance.