Chaning parameter roles (not using a operator)

frankie

Hi,

I find very inconvenient that I cannot change the roles of the variables in a dataset after it has been imported into RM. The only way --- as I understand --- is to re-import the data (takes a lot of time if I have many variables that need to be manually set) or to use an array of operators (clogs the view for starters). Is it not possible to reopen the dataset in the "import wizard" after the data has already been brought to RM? It would be so much convenient to change roles using the GUI.


BR,
Frankie

Andrew2

Hello Frankie,

You can use the Set Role operator. You can use this to change labels to regular and back again. You can also change between ids, predictions, clusters, weights and batches (whatever they are).

There's also a little known technique where you can just type free text in the target role field. I use 'ignore' and Rapidminer dutifully ignores the attribute. If you want to ignore multiple attributes, you have to use ignore1, ignore2 and so on.

regards

Andrew