A Receiver Operating Characteristic (ROC) curve is a graphical plot that illustrates the diagnostic ability of a binary classifier system as its discrimination threshold is varied. It provides a comprehensive, baseline-independent visualization of model performance by mapping the trade-offs between correctly identifying positive instances and incorrectly flagging negative ones. The Mechanics of an ROC Curve
The curve plots two vital metrics derived from a confusion matrix across all possible decision boundaries:
Y-Axis: True Positive Rate (TPR) – Also known as Sensitivity or Recall, this represents the proportion of actual positives correctly identified (
X-Axis: False Positive Rate (FPR) – Calculated as 1 – Specificity, this represents the proportion of actual negatives incorrectly flagged as positive (
As you adjust the decision threshold from strict (predicting positive only when 100% sure) to loose (predicting positive easily), the coordinates shift along the graph from the bottom-left corner (0,0) to the top-right corner (1,1). Interpreting the Plot & AUC
The overall quality of a classifier is quantified by the Area Under the Curve (AUC). AUC represents the probability that the model will rank a randomly chosen positive instance higher than a randomly chosen negative one. ROC-кривая – Википедия