MultiClassification
rk
New Altair Community Member
Dears,
I'm a newbie in the world of data mining and machine learning in general.. I'm trying to learn about it...
I'm trying to do multiclassification, but I have a few problems.
Suppose that I have 2 repositories, one with training set, and one with validation set. The training set contains 3 vectors for each classes, and there are 15 classes. My classes have for label 1,2, ...,15... While the validation set has 120 vectors in total.
The only method I'm able to use atm is knn, but when I try to use any other method I get errors like "method can not handle numeric labels", and the quick fixes show me that I need to discretizate them..
Could any1 care to explain me how to fix this?
Tks in advance fer your time !
PS> Sorry about my poor neglish skils.. English is not my 1st language..
I'm a newbie in the world of data mining and machine learning in general.. I'm trying to learn about it...
I'm trying to do multiclassification, but I have a few problems.
Suppose that I have 2 repositories, one with training set, and one with validation set. The training set contains 3 vectors for each classes, and there are 15 classes. My classes have for label 1,2, ...,15... While the validation set has 120 vectors in total.
The only method I'm able to use atm is knn, but when I try to use any other method I get errors like "method can not handle numeric labels", and the quick fixes show me that I need to discretizate them..
Could any1 care to explain me how to fix this?
Tks in advance fer your time !
PS> Sorry about my poor neglish skils.. English is not my 1st language..
Tagged:
0
Answers
-
Hello,
welcome to our community! The error message indicates that you classes are identified as being numerical (instead of nominal / categorical) by RapidMiner. If this is not desired, you could use an operator like "Numerical to Polynominal" on the label attribute in order to transform the values into true nominal classes.
By the way: please post RapidMiner specific questions like this one in the "Getting Started" or the "Problems" boards of this forum. I have moved this thread to "Getting Started".
Cheers,
Ingo0
