Knowledge Maxxing← Back to topic library
ML MaxxingBeginner75 sec
Confusion matrix
A confusion matrix is a table showing how many predictions of each class matched or did not match the true class, breaking results into correct and incorrect predictions of each type.
Prerequisite: Accuracy
What matters
Because it separates false positives from false negatives, a confusion matrix reveals what kind of mistakes a model is making, which a single accuracy number cannot show.
Example
A confusion matrix for a spam filter shows separately how many real spam emails were missed and how many real emails were wrongly flagged as spam.
Common misconception
A confusion matrix is not itself a single score; it is the underlying table that metrics like precision, recall and F1 are calculated from.
Recall question
What can a confusion matrix show that a single accuracy number cannot?
Keywords and related concepts
confusion matrixfalse positivefalse negativeprecision