Module kmeans

KMeans implementation for Apache Arrow Arrays.

Support l2, cosine and dot distances, see DistanceType.

Cosine distance are calculated by normalizing the vectors to unit length, and run l2 distance on the unit vectors.

Structs

• KMeans
  KMeans implementation for Apache Arrow Arrays.
• KMeansParams
  KMean Training Parameters

Enums

• KMeanInit
  KMean initialization method.

Functions

• compute_partition
• compute_partitions
  Compute partition ID of each vector in the KMeans.
• compute_partitions_arrow_array
  Compute partitions from Arrow FixedSizeListArray.
• kmeans_find_partitions
  KMeans finds N nearest partitions.
• kmeans_find_partitions_arrow_array