<![CDATA[Distributed Denial of Service (DDoS) attacks remain a major threat to network service availability due to their ability to generate massive traffic volumes that closely resemble legitimate activities. This study proposes an Explainable Ensemble Learning approach for DDoS detection using the CIC-DDoS2019 dataset. The proposed framework integrates Mutual Information-based feature selection to identify the 20 most relevant features, Synthetic Minority Over-sampling Technique (SMOTE) for class balancing, and a Voting Ensemble of Random Forest, XGBoost, and LightGBM classifiers. Model performance was evaluated using a 70:30 train-test split with Accuracy, Precision, Recall, F1-score, and ROC-AUC metrics. Experimental results achieved 99.80% Accuracy, 99.79% F1-score, and 0.9999 ROC-AUC. SHAP analysis identified Avg_Packet_Size and packet-length-related features as the most influential predictors, improving both detection performance and model interpretability. These findings demonstrate that integrating Ensemble Learning with Explainable Artificial Intelligence provides an accurate and transparent solution for DDoS detection interpretability.]]>
Copyrights © 2026
