Table 9 Performance of Hybrid Feature Selection with other feature selection/dimensionality reduction techniques
From: A novel time efficient learning-based approach for smart intrusion detection system
Feature selection/dimensionality reduction | Model building time | Prediction latency | Accuracy (%) | Recall (%) | Precision (%) | Specificity (%) | F-measure (%) |
|---|---|---|---|---|---|---|---|
Random Forests | 13.77661 | 1.443783 | 97.89 | 96.03 | 99.40 | 99.46 | 97.69 |
LightGBM | 11.12578 | 1.210907 | 97.81 | 96.12 | 99.40 | 99.45 | 97.73 |
Logistic Regression with L2 | 14.36824 | 1.44417 | 97.57 | 96.03 | 98.98 | 99.05 | 97.48 |
PCA | 7.376086 | 0.143057 | 97.82 | 96.12 | 99.40 | 99.44 | 97.73 |
Hybrid Feature Selection (RF + PCA) | 5.062662 | 0.138008 | 97.73 | 96.06 | 99.33 | 99.42 | 97.57 |