Agglomerative Hierarchical Clustering (AHC)

This is a Clustering method using Hierarchical Clustering Method, each point initially starts as a cluster, and slowly the nearest or similar most clusters merge to create one cluster. I.e. It sequentially merges similar clusters.

Tldr

It determines the distances in pairs, then merges them into clusters using linkage type.


Advantages:

  • There is no need to set the number of clusters.
  • With the right linkage, it can be used in Outlier Detection.
  • Using dendrograms, the results can be interpretable.

Disadvantages:

  • A good understanding of statistical properties of data is required to tune linkage type and distance metric.
  • It's challenging to optimize it.
  • Can be computationally expensive, especially for larger datasets.