Table 5 Performance of the predictive models for each serotype by using independent testing data. PCC was used as the feature selection method. SVM: Support Vector Machine; Sn: Sensitivity; Sp: Specificity; Acc: Accuracy; MCC: Matthews Correlation Coefficient
Serotype | Bin size of Peaks | Number of features | Classifiers | Sn | Sp | Acc | MCC |
|---|---|---|---|---|---|---|---|
I a | 9 | 28 | Random Forest | 66.0% | 61.4% | 61.9% | 0.168 |
SVM | 19.1% | 94.9% | 87.1% | 0.172 | |||
I b | 3 | 30 | Random Forest | 55.6% | 54.8% | 54.9% | 0.071 |
SVM | 54.0% | 38.9% | 41.0% | −0.050 | |||
III | 6 | 7 | Random Forest | 73.0% | 74.1% | 73.9% | 0.405 |
SVM | 68.0% | 71.3% | 70.6% | 0.336 | |||
V | 9 | 43 | Random Forest | 63.6% | 40.6% | 43.4% | 0.028 |
SVM | 5.5% | 94.1% | 83.4% | −0.007 | |||
VI | 7 | 18 | Random Forest | 70.4% | 70.3% | 70.4% | 0.381 |
SVM | 67.6% | 64.4% | 65.4% | 0.297 |