Module cluster_index 

K-means clustering index: Lloyd’s algorithm, cluster assignment, centroid tracking, cluster statistics, merge, split, ANN search by cluster probing (v1.1.0 round 13) K-means clustering index for approximate nearest-neighbour search.

Features:

• Lloyd’s algorithm with configurable k and maximum iterations
• Cluster assignment (nearest centroid)
• Centroid tracking (incremental updates as vectors are inserted)
• Cluster statistics (size, intra-cluster variance, centroid drift)
• Cluster merge (merge two closest clusters)
• Cluster split (split the largest cluster)
• ANN index search (probe nearest clusters for a query vector)

Structs

ClusterIndex

K-means clustering index supporting ANN search by cluster probing.

ClusterStats

Runtime statistics for a single cluster.

Enums

ClusterError

Errors produced by the cluster index.