Table 2. The Confusion matrix of estimated error rates with a 10-fold cross validation (CV) for feature genes.

(A) Out-of-bag (OOB) estimate of 5000 dierential expression genes is 51.85%. (B) Out-of-bag (OOB) estimate of 50 feature genes is 15.74%. Out-of-bag (OOB) estimate: also called out-of-bag (OOB) error, is a method for calculating prediction error of random forests. It uses bootstrap aggregating (bagging) to classify the sample data.

A. Out-of-bag (OOB) estimate of 5000 genes is 51.85%
	NC	MMA	SA	Error rate
NC	0	19	1	1
MMA	0	44	6	0.12
SA	0	30	8	0.79