Table 4.

Overall comparative results obtained from LOOCV of binary SVM classifier.

Metric	Performance without optimization			Performance with optimization
LOOCV accuracy	90%			92.5%
Confusion matrix		Pre	Post		Pre	Post
	Pre	18	2	Pre	18	2
	Post	2	18	Post	1	19
Specificity	0.90			0.95
Sensitivity	0.90			0.90
Area under the curve	0.9825			0.9850
Misclassification cost	1 (default)			0.0010
Kernel scale	1 (default)			0.0011

The rows of confusion matrix represent the actual class while the columns show the predicted class.