Research Article
A Machine Learning-Based Model for Stability Prediction of Decentralized Power Grid Linked with Renewable Energy Resources
Table 5
Performance of ML models on an undersampled dataset.
| Undersampling technique | Models | Accuracy (%) | Precision (%) | Recall (%) | -measure (%) | ROC |
| Cluster centroids | ANN | 94.4 | 94.6 | 94.1 | 94.4 | 98.7 | Averaged perceptron | 79.1 | 79.5 | 78.6 | 79 | 88.1 | Bayes point machine | 79.1 | 79.4 | 78.7 | 79 | 88.1 | Decision forest | 91.1 | 93.3 | 88.6 | 90.9 | 97.1 | Decision jungle | 88.9 | 88.8 | 89.1 | 88.9 | 95.8 | GBDT | 94.2 | 94.4 | 94.1 | 94.2 | 98.9 | LightGBM | 93.6 | 94.2 | 92.8 | 93.5 | 98.4 | Locally deep SVM | 90.9 | 91.1 | 90.6 | 90.9 | 97.2 | LR | 79.1 | 79.6 | 78.4 | 79 | 88.1 | SVM | 78.5 | 78.9 | 77.9 | 78.4 | 87.5 | XGBoost | 93.7 | 94.1 | 93.3 | 93.7 | 98.7 |
| Near miss | ANN | 92 | 73.2 | 89.8 | 91.9 | 98.1 | Averaged perceptron | 75.2 | 74 | 77.9 | 75.9 | 83.1 | Bayes point machine | 75.1 | 74.1 | 77.5 | 75.7 | 83 | Decision forest | 91.1 | 93.3 | 88.6 | 90.9 | 97.1 | Decision jungle | 85.3 | 83.7 | 87.7 | 85.7 | 93.1 | GBDT | 92.1 | 90.6 | 94 | 92.3 | 98.3 | LightGBM | 91.8 | 90.6 | 93.2 | 91.9 | 97.9 | Locally deep SVM | 88.3 | 88.3 | 88.3 | 88.3 | 94.6 | LR | 75.2 | 73.9 | 78 | 75.9 | 83 | SVM | 74.6 | 73.2 | 77.6 | 75.3 | 82.3 | XGBoost | 92.6 | 92.1 | 93.1 | 92.6 | 98.8 |
| Random undersampling | ANN | 94.5 | 97.8 | 91 | 94.3 | 99 | Averaged perceptron | 80.2 | 80.1 | 80.4 | 80.3 | 88.9 | Bayes point machine | 79.7 | 79.7 | 79.9 | 79.8 | 88.8 | Decision forest | 89 | 89.1 | 89 | 89 | 95.9 | Decision jungle | 89.2 | 87.8 | 91.1 | 89.4 | 95.9 | GBDT | 94.1 | 94.5 | 93.7 | 94.1 | 98.8 | LightGBM | 92.7 | 93.1 | 92.1 | 92.6 | 98.3 | Locally deep SVM | 90 | 89.8 | 90.2 | 90 | 96.6 | LR | 79.8 | 79.7 | 80 | 79.9 | 88.7 | SVM | 80.3 | 80 | 80.9 | 80.4 | 89 | XGBoost | 93.4 | 93.6 | 93.1 | 93.3 | 98.7 |
|
|