ROC curve from an array of Confusion Matrices (true positive rate and false positive rate)


Less is more. Stay pure. Stay poor.
It should work in the first figure if the two values are paired!

Of note, I use the terms SEN and 1-SEN, so I didn't look up the def for hit rate.
Last edited:


Ambassador to the humans
What exactly is your goal? A ROC curve is created from a single experiment. There probably is some meta analysis equivalent but you don't typically take independent experiments with pre selected cutoffs and create a ROC curve from that.

And you definitely don't independently sort the columns like you did in your second image.


Less is more. Stay pure. Stay poor.
I think you need to better describe your data in order for us to help. How do you get a SEN and SPEC out of a single observation subject, etc.?