I've set up a workflow whereas I train and test a SVR model. The feature data and the target data are normalized using the 'Normalizer' node and denormalized at the end of the workflow using the 'Denormalizer' node (model output of the Normalizer is connected to the model input of the Denormalizer).

The output data from the predictor ('LIBSVM_Class') however is not denormalized since it was not used as input data for the Normalizer node.

How can I denormalize the predicted values with the same parameters as the target column which was used for training the model?


I am not sure I understand your question, you can always normalize your data before going into the Learner and then denormalize it later, or simply use the original dataset. Sorry, I don't get it...

Hi there, 

You can denormalize the prediction column if you rename it to match the original target column name. I am not aware of any other way to do this.