New answers tagged

1

The ROC-AUC curves are used to find the best threshold that optimizes True Positive Rate vs False Positive Rate. Using it in a K-Fold cross-validation is a good practice to determine the best threshold to use. Then, your final test is here to validate that you did not overfit on some hyperparameters, including this threshold. So ROC-AUC must not be used ...


1

I made an experiment with a toy dataset in which first I trained a classifier with the original labels and plot the roc curve, then I switched the labels and train the same model on with the "new target" and even though the score in test set were the same, the roc auc plot looks slightly different. import numpy as np import matplotlib.pyplot as plt ...


Top 50 recent answers are included