Research Article
Improved Pre-miRNA Classification by Reducing the Effect of Class Imbalance
Table 7
The classification results of MiRNAClassify and three classification models over the merged datasets.
| Classification models | SE (%) | SP (%) | (%) |
| Human | SVM | 69.18 | 99.83 | 83.11 | SVM + SMOTE | 92.25 | 95.70 | 93.96 | Naive Bayes | 87.43 | 96.12 | 91.67 | Naive Bayes + SMOTE | 90.24 | 94.43 | 92.31 | Random Forest | 67.78 | 99.82 | 82.26 | Random Forest + SMOTE | 91.51 | 95.34 | 93.41 | MiRNAClassify | 97.93 | 98.30 | 98.11 |
| Animal | SVM | 69.03 | 98.14 | 82.31 | SVM + SMOTE | 91.61 | 94.85 | 93.21 | Naive Bayes | 85.04 | 95.03 | 89.90 | Naive Bayes + SMOTE | 90.83 | 92.61 | 91.71 | Random Forest | 69.52 | 98.72 | 82.84 | Random Forest + SMOTE | 91.12 | 95.01 | 93.05 | MiRNAClassify | 95.85 | 97.62 | 96.73 |
| Plant | SVM | 68.51 | 99.24 | 82.45 | SVM + SMOTE | 89.50 | 93.10 | 91.28 | Naive Bayes | 82.91 | 96.75 | 89.57 | Naive Bayes + SMOTE | 87.20 | 92.61 | 89.86 | Random Forest | 68.32 | 99.35 | 82.39 | Random Forest + SMOTE | 89.18 | 92.87 | 91.01 | MiRNAClassify | 93.37 | 97.91 | 95.61 |
|
|