Table 2.

Performance measurements for binary classification computed from a confusion matrix.

	Positive	Negative	Evaluation Focus
Predicted as positive	TP	FP	/
Predicted as negative	FN	TN	/
Acc	$\frac{\mathrm{TP}+\mathrm{TN}}{\mathrm{TP}+\mathrm{FP}+\mathrm{FN}+\mathrm{TN}}$		The overall efficiency of a classifier
Se (Recall)	$\frac{\mathrm{TP}}{\mathrm{TP}+\mathrm{FN}}$		The efficiency of a classifier to categorize positively labeled data
Sp	$\frac{\mathrm{TN}}{\mathrm{TN}+\mathrm{FP}}$		The efficiency of a classifier to categorize negatively labeled data
Precision	$\frac{\mathrm{TP}}{\mathrm{TP}+\mathrm{FP}}$		The data with positive labels correctly classified by the classifier

Note: Positive = normal, negative = pathological; TP = true positive, FP = false positive, FN = false negative, TN = true negative.