Thursday, 26 April 2018

ROC curve

ROC stands for 'Receiver Operating Characteristic'.It is a technique to visualize the performance of a binary classifier.

AUC, Area under curve will summarize the performance of a classifier in a single number.




ROC curve is a plot between True positive rate vs False positive rate at different threshold values.


True positive rate = no of true +ve vales predicted by classifier/ total no of +ve values


False positive rate= no of false +ve values predicted by classifier/ Total no of +ve values



Below shows a roc curve that represents ideal classifier




No comments:

Post a Comment

ROC curve

ROC stands for 'Receiver Operating Characteristic'.It is a technique to visualize the performance of a binary classifier. AUC, Ar...