INFO

Determines the optimal number of clusters by plotting within-cluster variance as a function of the number of clusters.

How It Works

  • Plot the sum of squared errors (SSE) for different values of
  • Look for the “elbow” point where SSE starts to level off
  • That point suggests the optimal number of clusters

What to Look For

  • Visual method, not a strict formula
  • Works best with convex clusters
  • Often used with K-Means

Application Models