Cluster

Cluster (cluster)

Assign cluster identifiers to numeric vectors.

Transform json

Minimal example

actions:
  - cluster: {}

JSON

{
  "actions": [
    {
      "cluster": {}
    }
  ]
}

Contents

• Minimal example

• Fields

• Schema

Fields

Field	Type	Required	Description
description	string		Describe this step.
condition	lua-expression (string)		Only run this action if the condition is met. Examples: 2 * count()
input-field	field (string)		Field containing the vector or numeric features for clustering. Examples: data_field
algorithm	Algorithm		Algorithm to use for clustering. Allowed values: kmeans, dbscan
k	number (integer)		Target number of clusters (k-means style algorithms). Examples: 42, 1.2e-10
epsilon	number (integer)		Distance threshold for density-based algorithms. Examples: 42, 1.2e-10
min-samples	number (integer)		Minimum neighbours for density-based clustering. Examples: 42, 1.2e-10
max-iterations	number (integer)		Maximum solver iterations. Examples: 42, 1.2e-10
output-field	field (string)		Field to write cluster identifiers into. Examples: data_field
emit-centroids	boolean (bool)		Emit cluster centroids alongside the assignments.
metric-field	field (string)		Optional field to capture distance or score metrics. Examples: data_field

Schema

• Algorithm Options

Algorithm Options

Value	Description
kmeans	Kmeans
dbscan	Dbscan