Comparison of Two Supervised Machine Learning In Intrusion Detection System

Intrusion Detection Systems (IDS) are critical components of network security designed to detect and prevent unauthorized access and malicious activity. Traditional rule-based IDSs are limited in their ability to adapt to evolving threats, so machine learning (ML) algorithms must be sought for intrusion detection. This paper presents a comparative analysis of IDSs using decision trees and random forest algorithms, focusing on their effectiveness, computational efficiency, and reliability. We investigate the implementation of decision tree-based models that offer interpretability and simplicity in rule generation, as well as ensembles of random forest trees, known for their excellent performance in handling complex datasets and reducing overfitting. Experimenting with the CICIDS2017 dataset, we evaluate performance metrics for both models, including precision, accuracy, recall, and F1 score. In addition, we analyze key