unsupervised
Hierarchical clustering
Explore nested group structure without fixing a cluster count in advance.
Requirements
- two or more scaled numeric features
- a defensible distance and linkage
Avoid when
- very large datasets
- mixed data types without an appropriate distance
Formula
d(A,B) is updated after each nearest-cluster merge
Interpretation
The dendrogram records the order and distance at which clusters merge.
Worked example
Inspect nested patient similarity and choose a cluster solution from a dendrogram.
Open example datasets