ROC Curve
Last Updated: July 29, 2026 | By Mihail Sebastian | AI Dictionary
A plot of a classifier's true positive rate against its false positive rate at every decision threshold, tracing the full menu of trade-offs it offers.
What is a ROC Curve?
A ROC curve plots a classification model’s true positive rate against its false positive rate at every possible decision threshold, tracing the full set of trade-offs between catching positives and raising false alarms.
The name, Receiver Operating Characteristic, is inherited from signal detection work on radar. In machine learning the curve does two jobs: comparing classifiers independently of any single threshold, and choosing the threshold to operate at.
How a ROC Curve Works
Most classifiers output a score, not a label; a threshold turns the score into a decision. Each threshold produces one confusion matrix, and from it one point on the plot: the true positive rate (recall) on the y-axis, the false positive rate on the x-axis. Sweeping the threshold from strict to lenient traces the curve.
A random guesser lands on the diagonal. The closer the curve bends toward the top-left corner (every positive caught, no false alarms), the better the model separates the classes.
The area under the curve (AUC) compresses the whole plot into one number: 0.5 for random guessing, 1.0 for a perfect ranking. AUC equals the probability that a randomly chosen positive case gets a higher score than a randomly chosen negative one. One caution: on heavily imbalanced data, ROC curves flatter the model, and a precision-recall view is more honest about false alarms.
Example of a ROC Curve
A hospital’s screening model scores 1,000 patients, 50 of whom have the disease. At a strict threshold it flags almost no one: barely any false alarms, but most sick patients missed. At a threshold of 0.5 it catches 40 of the 50 (true positive rate 0.8) while flagging 40 of the 950 healthy patients (false positive rate about 0.04). That is one point on the curve.
Lower the threshold further and both rates climb. The curve is the menu of available trade-offs; picking the operating point is a policy decision about which error the institution can better afford, not a modeling detail.
Related AI terms: Recall · Precision · Confusion Matrix · Model Evaluation · Classification
Did you like the ROC Curve gist?
Learn about 250+ need-to-know artificial intelligence terms in the AI Dictionary.
Mihail Sebastian — Writes about AI governance, regulation, and the technology behind them. Placeholder bio — replace with a real credential line. About