Receiver Operating Characteristics graphs, or ROC graphs, can be a useful approach to visualizing the benefits and costs that a classifier provides, especially when there are uncertain conditions. A ROC graph shows the entire space of performance possibilities by graphing true positive rates against false positive rates to identify if a classifier is better than random and is identifying the trade-offs between benefits and costs (Foster and Fawcett 2013, 215). When viewing a ROC graph, a classifier that produces results in the upper left of the dividing one to one line means that the classifier is utilizing information to make better predictions than random. ROC graphs are particularly good for classification, scoring, and class probability estimations. ROC graphs are a useful method of choice in displaying results when you are trying to focus primarily on the performance of different models specifically since they are indicating the different trade-offs each model is making.
Author: Logan Callen
Provost, Foster and Tom Fawcett. 2013. Data Science for Business. 2nd Edition. California: O’Reilly Media, Inc.