Table 4 Evaluation metrics report for all machine learning models.
From: Anomaly detection using machine learning and adopted digital twin concepts in radio environments
Class | Precision | Recall | F1 Score |
|---|---|---|---|
Random forest evaluation metrics report Overall accuracy: 0.98 | |||
0 (Normal) | 1.00 | 0.89 | 0.94 |
1 (Signal Drift) | 1.00 | 1.00 | 1.00 |
2 (Multipath Effect) | 1.00 | 0.98 | 0.99 |
3 (Localization Inaccuracy) | 1.00 | 0.98 | 0.99 |
4 (HShadLev) | 0.92 | 1.00 | 0.96 |
5 (LShadLev) | 0.99 | 0.97 | 0.98 |
KNN evaluation metrics report Overall accuracy: 0.81 | |||
0 (Normal) | 0.74 | 0.83 | 0.78 |
1 (Signal Drift) | 0.92 | 0.86 | 0.89 |
2 (Multipath Effect) | 0.99 | 0.88 | 0.93 |
3 (Localization Inaccuracy) | 0.95 | 0.80 | 0.87 |
4 (HShadLev) | 0.68 | 0.73 | 0.71 |
5 (LShadLev) | 0.61 | 0.72 | 0.66 |
Logistic regression evaluation metrics report Overall accuracy: 0.93 | |||
0 (Normal) | 0.73 | 0.97 | 0.83 |
1 (Signal Drift) | 1.00 | 1.00 | 1.00 |
2 (Multipath Effect) | 0.97 | 0.97 | 0.97 |
3 (Localization Inaccuracy) | 0.98 | 0.99 | 0.98 |
4 (HShadLev) | 0.92 | 0.77 | 0.84 |
5 (LShadLev) | 0.87 | 0.93 | 0.90 |
SVM evaluation metrics report Overall accuracy: 0.97 | |||
0 (Normal) | 0.99 | 0.93 | 0.96 |
1 (Signal Drift) | 1.00 | 1.00 | 1.00 |
2 (Multipath Effect) | 1.00 | 0.98 | 0.99 |
3 (Localization Inaccuracy) | 1.00 | 0.99 | 1.00 |
4 (HShadLev) | 0.91 | 0.97 | 0.94 |
5 (LShadLev) | 0.95 | 0.92 | 0.94 |
XGBoost evaluation metrics report Overall accuracy: 0.99 | |||
0 (Normal) | 1.00 | 1.00 | 1.00 |
1 (Signal Drift) | 1.00 | 1.00 | 1.00 |
2 (Multipath Effect) | 0.99 | 0.98 | 0.99 |
3 (Localization Inaccuracy) | 0.99 | 0.99 | 0.99 |
4 (HShadLev) | 0.98 | 1.00 | 0.99 |
5 (LShadLev) | 0.99 | 1.00 | 0.99 |
