How to use MATLAB for ROC analysis

ROC Curve Basics:

Update later

The ROC curve is determined by the Perfcurve function in the Statistics Toolkit.

The typical use is:

[X,Y,T,AUC] = Perfcurve (Labels,scores,posclass)

The output part X and Y represents the coordinates of the ROC curve, the AUC represents the area under the curve, T represents thresholds, when t=1 indicates there is a classification standard, can be 100% of all samples accurately classified, specificity and sensitivity is 100%. You can use plot (x, y) to get the simplest ROC curve.

The input section labels represents the real classification of the sample, scores represents the post-learning classification, and Posclass denotes the classification of the object we want to use. For example, when using ROC judgments for a model that predicts the gender of a skull, predicting the ROC discriminant curve for males and females is different, Posclass can be set to ' male ' or ' female '.

How to use MATLAB for ROC analysis