Need AUC for data which has label with polynomial value

varunmandalapu1

Hello,

I attached my data set with 205 Attributed in which Label is the last column named as Label. My label has 4 distinct values C1, C2, C3 & C4. I applied cross validation with Decision tree in training and Apply model and Performance in testing. The output has Accuracy and Kappa, but there is no AUC in the output. For AUC, Can I get AUC for polynomical value. Please suggest.

Thomas_Ott

Ah I read your post closer, to get AUC you need a max of two classes. You can't have four.

Thomas_Ott

Please post your XML using the code button above.

Usually this is related to the type of Performance operator you use.

varunmandalapu1

<输出/ >


<运营商激活= " true " class = "过程”兼容ibility="8.0.001" expanded="true" name="Process">

Telcontar120

If you want AUC, you will need to reformulate your existing polynominal label into a series of binominals: e.g., C1 vs not- C1, C2 vs not-C2, etc. Then you will get 4 different AUCs, one for each predicted class. Unfortunately RapidMiner won't do this automatically for you, but it is definitely possible to do on your own.

MartinLiebig

Hi,

you can build this with polynominal by binominal Classification and Remember-Recall.

Best,

Martin