"Keep Originals" option in Apply Feature set - Bug or Feature?

pblack476pblack476 Member Posts: 83 Maven
When selecting keep originals in that operator it brings back every attribute of the original ExampleSet. Even attributes I had previously filtered out with select attributes. I would like it to return me just the ones in the previous operator. Not the ones from the beginning of the process.

Either that is a bug or if it is a feature then I want some form of REMOVE attributes instead of SELECT because ther eis no such option as well and RM seems to recall the attributes even after I select them out.

Best Answer


  • Options
    sgenzersgenzer Administrator, Moderator, Employee, RapidMiner Certified Analyst, Community Manager, Member, University Professor, PM Moderator Posts: 2,959 Community Manager
    hi @pblack476 hmm not sure what you're trying to describe here. Can you please post some data and an XML process?
  • Options
    pblack476pblack476 Member Posts: 83 Maven
    edited November 2019

    Here it is. I am sorry for the poor clarity on my text:

    From my process you can see I have a big dataset with +300 Attributes. I Then select only one out of them and apply a new featureset to it. When I choose 'keep originals' I would expect it to return me the selected dataset that came INTO the apply FeatureSet operator. But instead it returns the ORIGINAL DATA from the beginning of the process. Completely ignoring my previous selection.

    If that is a Feature, I would then possibly request that the team consider adding a "Remove attributes" operator to make sure that the data going into the apply featureset operator does not contain the original dataset in it, or change the behaviour of the apply featureset 'keep originals' option. Which seems more reasonable.

  • Options
    pblack476pblack476 Member Posts: 83 Maven
    @varunm1 Well, then maybe I leave here a request for some kind of feature on future builds that allows me to do as I have described. Maybe a 'keep last' option on the operator itself.

Sign In or Register to comment.