Table 3 Evaluation metrics for... | Emerald Publishing

Table 3

Evaluation metrics for classification models

Evaluation metric	Purpose	Equation	Description
Precision	Single number summary of the precision recall curve	$\frac{T r u e p o s i t i v e s}{T r u e p o s i t i v e s + F a l s e p o s i t i v e s}$	True positives (TP) are the data-points that are predicted to be the positive class and it is correct; False positives (FP) are the data-points that are predicted to be the positive class and it is false
Recall	Ratio of true-positive predictions to the total actual positives; it measures the accuracy of positive predictions	$\frac{T r u e p o s i t i v e s}{T r u e p o s i t i v e s + F a l s e n e g a t i v e s}$	False negatives (FN) are the data-points that are the data-points that are predicted to be the negative class and it is false
MAUC	The averages of the AUC values of all pairs of classes in the dataset	$\frac{2}{C (C - 1)} \sum_{i < j} A (i . j)$	C is the number of classes in the dataset, and $A (i . j)$ is the AUC score between class $i$ and class $j$ ⁠. The MAUC score ranges from 0 to 1, and a perfect classifier has a MAUC score of 1
G-mean	A single value summary of a classifier's ability to identify correctly positive instances and negative instances	${(\prod_{i = 1}^{C} {R e c a l l}_{i})}^{\frac{1}{c}}$	C is the number of classes in the dataset
Mathews Correlation Coefficient (MCC)	The MCC score considers the confusion matrix values of all classes in the dataset	$\frac{T P \times T N - F P \times F N}{\sqrt{(T P + F P) (T P + F N) (T N + F P) (T N + F N)}}$	The MMC score ranges from −1 to 1, and a perfect classifier has an MMC score of 1
Mean Mathews Correlation Coefficient (MMCC)	An average of all the MCC values of all pairs of classes	$\frac{2}{C (C - 1)} \sum_{i < j} M C C (i . j)$	C is the number of classes in the dataset, and $M C C (i . j)$ is the MCC score between class $i$ and class $j$

Evaluation metric	Purpose	Equation	Description
Precision	Single number summary of the precision recall curve	$\frac{T r u e p o s i t i v e s}{T r u e p o s i t i v e s + F a l s e p o s i t i v e s}$	True positives (TP) are the data-points that are predicted to be the positive class and it is correct; False positives (FP) are the data-points that are predicted to be the positive class and it is false
Recall	Ratio of true-positive predictions to the total actual positives; it measures the accuracy of positive predictions	$\frac{T r u e p o s i t i v e s}{T r u e p o s i t i v e s + F a l s e n e g a t i v e s}$	False negatives (FN) are the data-points that are the data-points that are predicted to be the negative class and it is false
MAUC	The averages of the AUC values of all pairs of classes in the dataset	$\frac{2}{C (C - 1)} \sum_{i < j} A (i . j)$	C is the number of classes in the dataset, and $A (i . j)$ is the AUC score between class $i$ and class $j$ . The MAUC score ranges from 0 to 1, and a perfect classifier has a MAUC score of 1
G-mean	A single value summary of a classifier's ability to identify correctly positive instances and negative instances	${(\prod_{i = 1}^{C} {R e c a l l}_{i})}^{\frac{1}{c}}$	C is the number of classes in the dataset
Mathews Correlation Coefficient (MCC)	The MCC score considers the confusion matrix values of all classes in the dataset	$\frac{T P \times T N - F P \times F N}{\sqrt{(T P + F P) (T P + F N) (T N + F P) (T N + F N)}}$	The MMC score ranges from −1 to 1, and a perfect classifier has an MMC score of 1
Mean Mathews Correlation Coefficient (MMCC)	An average of all the MCC values of all pairs of classes	$\frac{2}{C (C - 1)} \sum_{i < j} M C C (i . j)$	C is the number of classes in the dataset, and $M C C (i . j)$ is the MCC score between class $i$ and class $j$