5
$\begingroup$

Can someone please explain the difference between AUC(Area under curve) and balanced accuracy in R?

For eg: In decision tree modelling I got the,

AUC : 0.91

balanced accuracy : 0.72

please explain how to interpret or understand this model based on the above two accuracy values?

$\endgroup$
6
$\begingroup$

AUC is not the same kind of measure as accuracy.

AUC is area under curve for the ROC chart and it has also a meaning in the rank order statistics.

http://en.wikipedia.org/wiki/Mann%E2%80%93Whitney_U_test#Area-under-curve_.28AUC.29_statistic_for_ROC_curves

Accuracy is simply a fraction of correctly predicted positives to all positives.

Suppose you have a data set with binary target variable where positive cases are 90% of all cases. Then you can simply classify everything to belong to positive cases and you will get accuracy of 90%.

Balanced accuracy is discussed here on page 2.

http://ong-home.my/papers/brodersen10post-balacc.pdf

I

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.