search

slychief / ismir2018_tutorial

120 stars 29 forks source link

The calculation of the AUC seems faulty #3

Open KarimMibrahim opened 5 years ago

KarimMibrahim commented 5 years ago

In "Part_1_Convolutional_Neural_Networks", when you calculate the AUC, you pass the test_pred to the roc_auc_score function, which is already rounded at 0.5.

Shouldn't you pass the probabilities "test_pred_prob"?

audiofeature commented 5 years ago

Dear Karim,

Thank you very much for the hint. You are right. We will fix this in the next version of this tutorial.

all the best, Thomas

Am 27.03.2019 um 17:29 schrieb Karim M. Ibrahim notifications@github.com:

In "Part_1_Convolutional_Neural_Networks", when you calculate the AUC, you pass the test_pred to the roc_auc_score function, which is already rounded at 0.5.

Shouldn't you pass the probabilities "test_pred_prob"?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/slychief/ismir2018_tutorial/issues/3, or mute the thread https://github.com/notifications/unsubscribe-auth/ALHE6c7NubG4CWZHoDf0a3MKK_o2wtcCks5va5xpgaJpZM4cOPrD.