I got Model accuracy = 0.879 and Model AUC = 0.932 (with Auc Roc)? Can someone explain to me why the two values are not the same; at the least why they aren't near?
Kind regards
Accuracy measures are dependent on the threshold chosen, whereas AUC is threshold-independent, so you should try checking the accuracy values at different threshold points. Furthermore, these values also depend on the ratio of positive to negative examples in the training dataset, and a highly imbalanced dataset can cause such discrepancies.
A simpler answer is Accuracy metric can show biased result for imbalanced datasets where ROC AUC metric
is designed to overcome that limitation. So you should focus on ROC AUC score cause thats the one that counts
Thank you for all body;
I have discovered that I was mistaken. I have made some mistakes in my program.
Best regard.
- Badges
- Science method