🎉 🎉 RAPIDMINER 9.10 IS OUT!!! 🎉🎉

Download the latest version helping analytics teams accelerate time-to-value for streaming and IIOT use cases.

CLICK HERE TO DOWNLOAD

Scale and center 'label' attribute for prediction?

ben_hben_h Member Posts: 17 Contributor II
edited November 2018 in Help
I need some help to set my input data correctly for a cross-validation task. I have only continuous numeric variables, and the label or target variable is also a continuous numeric variable.

I separate my data set into test and training sets, then train using a (neural network) operator. I have the Apply Model operator for the test set, and use the Performance operator for evaluation.

As my variables/attributes are output from different processes and have very different ranges, I need to scale them and I also centre them for use in ANN (SOM in particular requires this). Is it necessary to also scale and centre my label/target attribute?

(edit: simplified to remove a second question)

Answers

  • MariusHelfMariusHelf RapidMiner Certified Expert, Member Posts: 1,869   Unicorn
    No, that is not necessary. The SOM operator ignores the label, and the Neural Net does not care about the scale of the label.

    Best regards,
    Marius
Sign In or Register to comment.