Table 3 Performance metrics used for model evaluation across folds and repetitions.

From: Integrating machine learning and time-to-event models to explain and predict risk of hospitalization due to dengue in Colombia

Metric

Definition

Interpretation

Accuracy

\(\dfrac{\text {TP + TN}}{\text {TP + TN + FP + FN}}\)

Overall proportion of correct classifications.

Positive predictive value (PPV, also called precision)

\(\dfrac{\text {TP}}{\text {TP + FP}}\)

Probability that a predicted positive is truly positive.

Negative predictive value (NPV)

\(\dfrac{\text {TN}}{\text {TN + FN}}\)

Probability that a predicted negative is truly negative.

Sensitivity (also called recall)

\(\dfrac{\text {TP}}{\text {TP + FN}}\)

Proportion of true hospitalized patients correctly classified.

Specificity

\(\dfrac{\text {TN}}{\text {TN + FP}}\)

Accuracy in identifying non-hospitalized individuals.

F1-score

\(2 \times \left( \dfrac{\text {PPV} \times \text {sensitivity}}{\text {PPV} + \text {sensitivity}}\right)\)

Balances overfitting positives and missing true positives.

Back to article page