MuhammadLab
LearningAlgorithmBeginner-friendly

K-Means — Clustering (Explained)

Learn K-means clustering: centroids, assignments, inertia, and how to pick k.

What you'll learn

  • How K-means alternates between assignment and centroid update.
  • Why scaling and initialization matter.
  • How the “elbow method” helps pick k.

Goal: group points into k clusters

K-means chooses k centroids and assigns each point to the nearest centroid.

Then it updates centroids to be the mean of their assigned points and repeats until stable.

Why it can fail

K-means prefers spherical, similarly-sized clusters because it relies on Euclidean distance to centroids.

Bad initialization can lead to poor local minima; K-means++ helps a lot.

Picking k

The elbow method looks at inertia (within-cluster sum of squares) as k increases.

Choose k where extra clusters provide diminishing returns.

Key takeaways

  • Scale features before K-means.
  • Use K-means++ initialization if available.
  • Best for roughly spherical clusters.
  • Elbow method is a heuristic, not a guarantee.

Want more ML topics added here (SVM, Naive Bayes, CNN, PCA, Decision Trees)?

Browse Machine Learning ->