What Are Accuracy, Precision, Recall, and F1 Score in Machine Learning?

• French B
• Geography
• German A
• German A Lit
• German ab initio
• German B
• Global Politics
• History
• History (First Exam 2028)
• Mathematics Analysis and Approaches (AA)
• Mathematics Applications & Interpretation (AI)

• Philosophy
• Physics
• Psychology
• Psychology (First Exam 2027)
• Spanish A
• Spanish A Lit
• Spanish ab initio
• Spanish B
• Sports, exercise and health science (SEHS)
• Theory of Knowledge (TOK)
• World Religions

• MYP English Lang & Lit
• MYP Extended Mathematics
• MYP History

• MYP Individuals & Societies
• MYP Integrated Sciences
• MYP Physics

• MYP Standard Mathematics

Suppose a medical classifier produces $TP=40$, $TN=50$, $FP=10$, and $FN=20$. Its accuracy is $\frac{90}{120}=0.75$, precision is $\frac{40}{50}=0.80$, recall is $\frac{40}{60}\approx0.67$, and F1 score is approximately $0.73$.

A common misconception is that high accuracy always means a good model. With an imbalanced dataset, a model could predict the majority class every time and achieve high accuracy while failing to identify important minority-class cases. Recall is especially important when false negatives are costly, while precision matters when false positives are costly.

IB Exam Technique

For A4.3 Machine learning approaches (HL only), define each metric, show the correct formula, substitute values from the confusion matrix, and interpret the result in context. When asked to evaluate a classifier, explain why accuracy alone may be misleading and justify which metric is most appropriate.