Performance Metrics
Performance metrics tell you something about the performance of a machine learning model. Each metric has a specific focus. Because of the confusion matrix’ nature, a lot of metrics have a close sibling. Equally confusing is that many performance metrics have multiple synonyms, depending on the context.
Given true
The following performance metrics are independent of the prevalence.
- True Positive Rate (Recall, Sensitivity, Power) vs. True Negative Rate (Specificity, Selectivity)
- False Positive Rate (Fall-out) vs. False Negative Rate (Miss Rate)
Given predicted
The following performance metrics are dependent of the prevalence.
- Positive Predictive Value (Precision) vs. Negative Predictive Value
- False Discovery Rate vs. False Omission Rate
General performance
General performance metrics evaluate the model without zooming in on specific true or predicted values.
- Prevalence
- Accuracy vs. Misclassification Rate
- Balanced Accuracy vs. Balanced Misclassification Rate
- F1 Score
- Jaccard Index
- Classification Success Index
- Matthews Correlation Coefficient
- Kulczynski’s Measure
- Ground Truth Index
- Fowlkes-Mallows Index
- Positive Likelihood Ratio
- Negative Likelihood Ratio
- Diagnostic Odds Ratio
- Informedness
- Markedness
- Optimization Precision
- Cohen’s Kappa
- Youden’s Index
- K-Index
- Discriminant Power
- Uncertainty Coefficient
- Null Error Rate
- ROC Curve
- Area Under the Curve (AUC)