Tuesday, 25 February 2014

"ROC" curves for science

ROC curves plot true positive rates (TPR) versus false positive rates (FPR) for a given classifying procedure.  This post discuses these curves as they might apply to different scientific disciplines.

It suggests that such curves are a more sophisticated way of judging the trade-offs inherent in science, and briefly discusses ways to improve ("lower") these curves, including awareness of "researcher degrees of freedom".

No comments:

Post a Comment