"Loop Labels removing label"

rexeg

Hello I am working with a dataset with polynominal data. Using nominal to binominal I generated 4 attributes for the original label attribute.  I would like to create models for each of these label attributes.  I referenced the 'loop labels' documentation, used set role to set one to 'label'  and the rest to 'label2', 'label3', and so on.  After dragging exa to 'loop labels' with my sub process inside, I get the message "input example set must have special attribute label". This is weird because I have set the roles of 4 attributes as label.  Checking the focus inside the loop labels operator, I discover that the attribute I designated as 'label' is missing and the other three (label2, label3, label4) remain.  Help would be appreciated. I have inserted images illustrating this issue.

 

All labels have been setInside loop labels, label has been removed

 

Edit: I would also like to point out that the built in tutorial for 'loop labels' also experiences this error.

jczogalla

Hello @rexeg,

RapidMiner only has one special role that is "the" label. Loop Labels uses that mechanic to rotate the label around for the specified label1, label2... custom attribute roles.

The error that you are getting looks like just a setup error, i.e. you should be able to still run the process and have the desired result. If you look at the tutorial process of Loop Labels, you can see that the same setup error occurs there and it still works.

The reason is that not all operators can precalculate the meta data completely, leading to some setup errors that will not prevent a process from running.

 

I hope this helps

Jan