RapidMiner

‎02-01-2018 10:31 AM

 In RapidMiner Studio it is often useful to have certain attributes be "ignored" by modelers, or just have more than one "label" special attribute. Here's a hidden trick to make this happen using the "Titanic" data set:

 

1. Using the Set Role operator, delete the word "regular" in the target role field and replace it with a name that is not part of the standard list (regular, label, id, etc...).

 

Screen Shot 2018-02-01 at 10.15.04 AM.png

 

2. If you want to add additional "roles", go to "set additional roles" and do the same thing:

 

Screen Shot 2018-02-01 at 10.22.23 AM.png

 

That's about it. You can see here that the Decision Tree shows Sex and Age in the first tree, and ignores it in the second.

 

Screen Shot 2018-02-01 at 10.24.10 AM.pngTitanic data set with no custom roles               Screen Shot 2018-02-01 at 10.24.38 AM.pngTitanic Decision Tree with no custom roles. Note "Sex" at top node.

 

Screen Shot 2018-02-01 at 10.26.14 AM.pngTitanic data set with two custom roles for Sex and Age         Screen Shot 2018-02-01 at 10.27.09 AM.pngsame data set showing the names of the custom roles in the Statistics view

 

 

Screen Shot 2018-02-01 at 10.26.23 AM.pngTitanic Decision Tree with custom roles being ignored.

 

The process XML is attached for reference.

 

Scott

 

 

Scott Genzer
Senior Community Manager
RapidMiner, Inc.
Comments
Unicorn
Unicorn

I use this trick all the time, it is very helpful!