Table 2.

Performance measures of the random forest classifier on the training dataset of 172 Met residues.

	Experimental ‘Liable’	Experimental ‘Non-liable’
Predicted ‘Liable’	15 (TP)	3 (FP)	Precision = 0.83 TP/(TP+ FP)
Predicted ‘Non-liable’	3 (FN)	151 (TN)
	Recall/Sensitivity = 0.83 TP/(TP+ FN)	Specificity = 0.98 TN/(TN+ FP)	Accuracy = 0.96 (TP+ TN)/Total
Matthew’s Correlation Coefficient (MCC) = $y_0=Y\lambda -s^{+}$ = 0.81

TP = True Positive, TN = True Negative, FP = False Positive, FN = False Negative

All predictions are ‘out-of-bag’ (OOB); that is predictions on each data point were made only using the trees not generated using that point.