Confusion matrix
A table that counts a classifier’s right and wrong decisions by type: true and false positives, true and false negatives.
For a yes-or-no task it has four cells. True positive: flagged, and should be. False positive: flagged, but should not be, a false alarm. False negative: missed. True negative: correctly left alone. Precision, recall and accuracy are all computed from these four counts. Accuracy alone misleads on rare events: a model that never flags fraud is 99% accurate when 1% of transactions are fraud.
Related terms
- Precision (AI metric): The share of returned results that are actually correct: of everything the system flagged or retrieved, how much was right.
- Recall (AI metric): The share of all correct items that the system actually found: of everything it should have caught, how much it did.
- F1 score: One number that combines precision and recall as their harmonic mean, so it is high only when both are high.
- Ground truth: The known correct answers or labels that a model’s output is compared against in an evaluation.
