confusion matrix

Confusion Matrix

A table showing the four outcomes of a binary classifier: true positives, false positives, true negatives, and false negatives — the foundation for evaluating classification model performance.

When you'd see it: Machine learning model evaluation for any binary classification problem — fraud detection, churn prediction, spam filtering, medical diagnosis. The confusion matrix is the source of precision, recall, F1 score, and accuracy.

Why it matters: Accuracy alone is misleading for imbalanced classes. A model that predicts 'not fraud' for every transaction has 99.9% accuracy on a dataset where 0.1% are fraud — and catches zero fraud. The confusion matrix makes visible exactly where the model is failing.

Common mistakes: Optimizing for accuracy when classes are imbalanced. Almost any model achieves high accuracy by predicting the majority class. Always look at precision, recall, and F1 — not just accuracy — for imbalanced classification problems.

Related precision, recall, F1 score, AUC, logistic regression