<![CDATA[The security of university networks faces an increase in attack surfaces due to the high dependence on digital services. Machine learning-based Intrusion Detection System (IDS) is one of the approaches to recognize benign and malicious traffic patterns. This study analyzes the benchmark machine learning model in the CIC-IDS2017 dataset through a literature study approach and comparative analysis. The models compared include Decision Tree, Random Forest, Support Vector Machine, Naive Bayes, Artificial Neural Network, as well as additional benchmarks XGBoost, KNN, and Logistic Regression. Evaluation parameters include accuracy, precision, recall, F1-score, false positive, false negative, and class distribution. The benchmark results showed that Random Forest achieved an accuracy of 99.67% with precision, recall, and malicious class F1-scores of 99.90%, 99.67%, and 99.67%, respectively in one of the published configurations. Another comparative study reported that XGBoost and Random Forest achieved 99.96% accuracy in their experimental configurations. Analysis of the class distribution showed a strong imbalance so that accuracy was not sufficient to assess the effectiveness of the IDS. The study recommends macro-F1 evaluation, per-class recall, false negative, and testing on more representative data before the model is used in a college production environment.]]>
Copyrights © 2026
