rohit.vision

HDBSCAN

ml ml unsupervised clustering hdbscan density 1 min read

Hierarchical density-based clustering that handles varying densities

Overview

  • Only needs $n$ (min samples) parameter
  • Handles varying densities
  • Slower than k-means but more versatile

When to Use

  • Recommended as first clustering approach
  • Best for unknown number of clusters with varying densities