Numerical Hard SOM with Hardclustering

Run Numerical Hard SOM on regular vector collections to learn a SOM model and apply it to predict the input data corresponding HardClustering.

(numerical_hard_som_with_hardclustering_hyperparameters_workflow_to_scala_engine)=

(numerical_hard_som_with_hardclustering_hyperparameters)=

Hyper-parameters

• restrictedToColumns : String JSON Array of column names containing numerical features from which learning.
• maxIterations : Maximum number of SOM iterations.
• width : SOM grid width.
• length : SOM grid length.
• tMin : Minimal temperature from which starts iterations.
• tMax : Maximal temperature from which ends iterations.
• initializedModel : It is a String which can takes two kind of values :
  • The dataLocationId of an existing model if the user want a custom initialization.
  • An empty String to use the default initialization.
• persistence : String taken in enumeration of Spark memory level available. vector will be made.