Calibration Curve
A graph that compares predicted probabilities with actual outcomes to evaluate how well a machine learning model's confidence matches reality.
What is a Calibration Curve?
A calibration curve compares predicted probabilities with the proportion of outcomes that actually occur. For example, if a well-calibrated model assigns a 70% probability to many events, roughly 70% of those events should occur. The closer the curve is to perfect calibration, the more accurately the model’s probability estimates reflect real-world outcomes.
Why is a Calibration Curve Important?
A model can classify outcomes accurately while still producing unreliable probability estimates. Calibration curves help teams determine whether prediction confidence can be trusted. This is particularly important when probabilities influence risk assessments, thresholds, resource allocation, or other important decisions.
Common use cases
Calibration curves are commonly used in healthcare prediction, credit risk assessment, fraud detection, insurance, classification models, and other probability-based decision systems.