|
| Proposed solutions | Performance metrics | Limitations |
|
| EBT [3] | Sensitivity, specificity, false-positive rate, and -score | Increased computational time |
| CNN and LSTM [20] | -score, area under curve (AUC), precision, and recall | No normalization and parameter tuning |
| PCA and RE [21] | ROC, specificity, and sensitivity | PCA only works for linear data |
| Fuzzy logic [22] | Generalized bell curve membership function | Increased computational time |
| Fuzzy logic [23] | Accuracy, -score, AUC | Issues related to renewable sources are not handled |
| Semisupervised deep neural network (DNN) [25] | Precision, true and false-positive rates, recall, and -score | High false-positive rate |
| LSTM and GMM [26] | AUC, MCC, recall, and accuracy | Data imbalance is not handled |
| MODWPT and RUSBoost [27] | -score, AUC, and precision | Oversampling issue is not tackled |
| Blackhole algorithm [28] | Average execution time and convergence | High false-positive rate |
| MIC and CFSFDP [29] | -score, precision, and recall | Low precision and recall |
| LSTM [31] | Accuracy, sensitivity, and specificity | Overfitting is not handled well |
| LSTM and regression [32] | -score, recall, and precision | Low -score |
|
