Confusion Matrix

What Confusion Matrix is

The matrix breaks predictions down per class so you can see whether errors cluster in one direction, which a single accuracy figure hides completely.

How it works

Counts of correct and incorrect predictions per class are tabulated on the test set; precision, recall, specificity and F1 are all derived from those four cells in the binary case.

Why it matters

It is the fastest way to explain a classifier's behaviour to a non-technical stakeholder who cares about a specific error type.

Common uses

• →Medical screening review
• →Fraud model tuning
• →Moderation system audits