Smart network intrusion detection system using stacked ensemble learning and data analytics

The growing prevalence of cyberattacks poses a significant threat to the confidentiality and integrity of sensitive data transmitted over networks. Effective Network Intrusion Detection Systems (NIDS) are essential to detect and mitigate these threats. This study introduces a novel model, Network Intrusion Detection System using Stacked Ensemble Learning Technique (NIDSSELT), which addresses key challenges in existing NIDS solutions, including class imbalance, detection accuracy, and false alarm rates. The proposed NIDSSELT framework integrates Extra Trees (ET) and Xtreme Gradient Boosting (XGB) as base classifiers with Logistic Regression (LR) as the meta-classifier, leveraging a Stacked Ensemble Learning Technique for robust predictions. To address class imbalance, the model employs K-means SMOTE for oversampling and representation of minority classes. Comprehensive evaluation using accuracy and F1-score, validated through 10-fold cross-validation, demonstrates the superior performance of NIDSSELT compared to existing single-classifier and ensemble-based models, offering a promising solution for enhanced network security.