Operator ComputeSVMError

A special evaluation operator used for obtaining some results for the regression SVM. Values in TheTargetValueAttribute are compared to those in ThePredictedValueAttribute. The average loss is determined taking the asymmetric loss function into account. That is why the SVM parameters are needed here as well. Note that they must have the same value as for the operator SupportVectorMachineForRegression, which must have preceded this evaluation operator in the chain.

Parameter

Parameter Object Type optional min_arg max_arg Remarks
TheInputConcept Concept Input no 1 1 inherited
TheTargetValueAttribute BaseAttribute Input no 1 1 actual values
ThePredictedValueAttribute BaseAttribute Input no 1 1 predicted values
LossFunctionPos Value Input yes 0 1 (same values
LossFunctionNeg Value Input yes 0 1 as in SVM-
C Value Input no 1 1
Epsilon Value Input no 1 1 ForRegression)