DataTransformation (power)

Oprick

Hello,

I feel a bit stupid asking this... After 2 years using RM.
I would like to perform a power transformation like yeo-johnson and I'm not finding an operator to do that.

Does it exist?

Regards

David_A

Hi @Oprick ,

currently there is no operator for performing power transformations, at least that I'm aware of. In case of the simpler ones like Box-Cox or Yeo-Johnson, you could rebuild them with the expression parser in Generate Attributes operator. As it's a quite handy transformation, we have already planed to add to an upcoming release of the TimeSeries functionality.

Best,

David

hughesfleming68

Hi @Oprick,

The quickest way is to use use Execute Python and sklearn. It is part of sklearn preprocessing.

https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.PowerTransformer.html

David_A

Hi @Oprick ,

currently there is no operator for performing power transformations, at least that I'm aware of. In case of the simpler ones like Box-Cox or Yeo-Johnson, you could rebuild them with the expression parser in Generate Attributes operator. As it's a quite handy transformation, we have already planed to add to an upcoming release of the TimeSeries functionality.

Best,

David

Oprick

Hi guys,
I currently use caret (R) and I perform this task with Execute R.
I just thought that somewhere in RM would exist an operator for this

Thanks you both for your very quick replies.
Regards