the.com/k means
Unsupervised learning that sorts messy data into k groups by finding their natural centers.
means An algorithm that divides a dataset into k clusters by repeatedly repositioning group centers until data points settle into stable neighborhoods around them.
from Developed in 1957 by Stuart Lloyd at Bell Labs, formalized and published in 1982. The name describes exactly what it does: find k means (averages) that minimize scattered chaos. Simple enough that reinventing it became a rite of passage for mathematicians.
initialization matterspoor starting positions can trap algorithm in weak local solutions
convergence guaranteedalways finds a solution, though not necessarily the global best one
the k problemchoosing the right number of clusters remains annoyingly unsolved in practice
speed advantagescales well to massive datasets compared to hierarchical alternatives