Pages: [1]
  Print  
Author Topic: Newb question: grouping selected nominal values  (Read 1350 times)
ctagg
Newbie
*
Posts: 1


« on: September 03, 2008, 03:49:59 PM »

New to RapidMiner and Data Mining in general, so apologies if this is a stupid question.

I have a data set with a type attribute and there are three main types and about 15 minor types.

I want to rename nearly all the minor types to 'other'. I've managed to do this using the AttributeValueMapper (though it seems a bit cumbersome, having to an AttributeValueMapper for each of the types -- is there a better way?). However, when I come to plot the options I run into difficulties, particularly with the legend, getting the error:

"Plotter: cannot draw nominal legend since number of different values is too high (more than 20)! Using numerical legend instead." The problem seems to be that it wants to plot the types that have been replaced by 'other

How do I 'lose' the possible nominal values that have now been converted to 'other'? Is it possible?

Thanks
Chris
Logged
Tobias Malbrecht
Global Moderator
Sr. Member
*****
Posts: 293



WWW
« Reply #1 on: September 06, 2008, 09:09:06 PM »

Hi Chris,

I want to rename nearly all the minor types to 'other'. I've managed to do this using the AttributeValueMapper (though it seems a bit cumbersome, having to an AttributeValueMapper for each of the types -- is there a better way?).

We recently extended the AttributeValueMapper. It is now capable of handling multiple mappings at once, i.e. you only need one AttributeValueMapper to replace different values.

How do I 'lose' the possible nominal values that have now been converted to 'other'? Is it possible?

Actually, the extended AttributeValueMapper accomplishes this as well. To use the extended operator, you simply have to check out the current developer version from our CVS. You will find an explanation on how to do that using the Eclipse IDE at

http://rapid-i.com/content/view/25/48/

Regards,
Tobias
Logged

Tobias Malbrecht
Director of Product Marketing
RapidMiner
Pages: [1]
  Print  
 
Jump to: