I think it's the proportion of events in the training data that changes as you move along the ROC curve. The more events you have, the better your model can discriminate between positive and negative cases.
The ROC curve shows the trade-off between the true positive rate and the false positive rate, so as you move along the curve, the probability cutoff for scoring must be changing. I'm pretty sure that's the right answer.
upvoted 0 times
...
Log in to Pass4Success
Sign in:
Report Comment
Is the comment made by USERNAME spam or abusive?
Commenting
In order to participate in the comments you need to be logged-in.
You can sign-up or
login
Hortencia
4 days agoDoug
5 days agoNa
5 days agoDonte
7 days agoYoulanda
9 days ago