calc the area under the curve (ROC CURVE)?

KNIME Analytics Platform
1

Hello

I have a Joiner Node with 

(TruePositives    FalsePositives    TrueNegatives    FalseNegatives    Recall    Precision    Sensitivity    Specifity    F-measure...)

How one can calc the area under the curve (ROC CURVE)?

Thanks

Malik

2

The ROC curves node outputs the AUC in its output table.

(note, of course you will need the classification probabilities for calculating the ROC curve, which are not mentioned in your post).

Regards,
Philipp

3

Hi malik,

would it be an option to calculate the AUC with a KNIME scripting node and use the output for your needs?

This is a good reference on how to do it: http://stats.stackexchange.com/questions/145566/how-to-calculate-area-under-the-curve-auc-or-the-c-statistic-by-hand

You can use a Java node or an R snippet node. You can also use one of the many available R libraries provided they have the ROC/AUC function you need.

Cheers,
Marco.