A MultiStep operator (creates several ColumnSets for the output Concept). The input Concept is segmented according to the clustering method KMeans (an external learning algorithm). The number of ColumnSets in the output concept is therefore not known before the application of this operator. However, the parameter HowManyPartitions specifies a maximum for this number. The parameter OptimizePartitionNum is a boolean that specifies if this number should be optimized by the learning algorithm (but it will not exceed the maximum). The parameter SampleSize gives a maximum number of learning examples for the external algorithm. The algorithm (KMeans) uses ThePredictingAttributes for clustering; these attributes must belong to TheInputConcept.

ParameterName ObjType Type Remarks
TheInputConcept CON IN inherited
HowManyPartitions V IN positive integer
OptimizePartitionNum V IN true or false
ThePredictingAttributes BA List IN  
SampleSize V IN positive integer
TheOutputConcept CON OUT inherited