- condition positive (P)
- the number of real positive cases in the data
- condition negative (N)
- the number of real negative cases in the data
- true positive (TP)
- eqv. with hit
- true negative (TN)
- eqv. with correct rejection
- false positive (FP)
- eqv. with false alarm, Type I error
- false negative (FN)
- eqv. with miss, Type II error
- sensitivity, recall, hit rate, or true positive rate (TPR)
- [math]\displaystyle{ \mathrm{TPR} = \frac {\mathrm{TP}} {\mathrm{P}} = \frac {\mathrm{TP}} {\mathrm{TP}+\mathrm{FN}}= 1 - \mathrm{FNR} }[/math]
- specificity, selectivity or true negative rate (TNR)
- [math]\displaystyle{ \mathrm{TNR} = \frac {\mathrm{TN}} {\mathrm{N}} = \frac {\mathrm{TN}} {\mathrm{TN} + \mathrm{FP}} = 1 - \mathrm{FPR} }[/math]
- precision or positive predictive value (PPV)
- [math]\displaystyle{ \mathrm{PPV} = \frac {\mathrm{TP}} {\mathrm{TP} + \mathrm{FP}} = 1 - \mathrm{FDR} }[/math]
- negative predictive value (NPV)
- [math]\displaystyle{ \mathrm{NPV} = \frac {\mathrm{TN}} {\mathrm{TN} + \mathrm{FN}} = 1 - \mathrm{FOR} }[/math]
- miss rate or false negative rate (FNR)
- [math]\displaystyle{ \mathrm{FNR} = \frac {\mathrm{FN}} {\mathrm{P}} = \frac {\mathrm{FN}} {\mathrm{FN} + \mathrm{TP}} = 1 - \mathrm{TPR} }[/math]
- fall-out or false positive rate (FPR)
- [math]\displaystyle{ \mathrm{FPR} = \frac {\mathrm{FP}} {\mathrm{N}} = \frac {\mathrm{FP}} {\mathrm{FP} + \mathrm{TN}} = 1 - \mathrm{TNR} }[/math]
- false discovery rate (FDR)
- [math]\displaystyle{ \mathrm{FDR} = \frac {\mathrm{FP}} {\mathrm{FP} + \mathrm{TP}} = 1 - \mathrm{PPV} }[/math]
- false omission rate (FOR)
- [math]\displaystyle{ \mathrm{FOR} = \frac {\mathrm{FN}} {\mathrm{FN} + \mathrm{TN}} = 1 - \mathrm{NPV} }[/math]
- Threat score (TS) or Critical Success Index (CSI)
- [math]\displaystyle{ \mathrm{TS} = \frac{\mathrm{TP}}{\mathrm{TP} + \mathrm{FN} + \mathrm{FP}} }[/math]
- accuracy (ACC)
- [math]\displaystyle{ \mathrm{ACC} = \frac {\mathrm{TP} + \mathrm{TN}} {\mathrm{P} + \mathrm{N}} = \frac {\mathrm{TP} + \mathrm{TN}} {\mathrm{TP} + \mathrm{TN} + \mathrm{FP} + \mathrm{FN}} }[/math]
- balanced accuracy (BA)
- [math]\displaystyle{ \mathrm{BA} = \frac {TPR + TNR}{2} }[/math]
- F1 score
- is the harmonic mean of precision and sensitivity
- [math]\displaystyle{ \mathrm{F}_1 = 2 \cdot \frac {\mathrm{PPV} \cdot \mathrm{TPR}} {\mathrm{PPV} + \mathrm{TPR}} = \frac {2 \mathrm{TP}} {2 \mathrm{TP} + \mathrm{FP} + \mathrm{FN}} }[/math]
- Matthews correlation coefficient (MCC)
- [math]\displaystyle{ \mathrm{MCC} = \frac{ \mathrm{TP} \times \mathrm{TN} - \mathrm{FP} \times \mathrm{FN} } {\sqrt{ (\mathrm{TP}+\mathrm{FP}) ( \mathrm{TP} + \mathrm{FN} ) ( \mathrm{TN} + \mathrm{FP} ) ( \mathrm{TN} + \mathrm{FN} ) } } }[/math]
- Informedness or Bookmaker Informedness (BM)
- [math]\displaystyle{ \mathrm{BM} = \mathrm{TPR} + \mathrm{TNR} - 1 }[/math]
- Markedness (MK)
- [math]\displaystyle{ \mathrm{MK} = \mathrm{PPV} + \mathrm{NPV} - 1 }[/math]
Sources: Fawcett (2006),[1] Powers (2011),[2] Ting (2011),[3] and CAWCR[4] Chicco & Jurman (2020)[5].
|