Performance Measure

In binary classification problems, samples can be treated as:

• True Positive (TP)
• False Positive (FP)
• True Negative (TN)
• False Negative (FN)

Especially, TP + FP + TN + FN = all samples.

In order to be more intuitive, we get a table:

		Predicted class	Predicted class
		Class = 1	Class = 0
Actual class	Class = 1	TP (11)	FN(10)
Actual class	Class = 0	FP(01)	TN(00)

Precision = TP / ( TP + FP)

Sensitivity / Recall = TP / (TP + FN)

Specificity = TN / (TN + FP)