BEHAVIORAL ANALYSIS OF MALICIOUS ACTIVITIES IN NETWORK TRAFFIC USING MACHINE LEARNING ALGORITHMS
Prof. Naved Raza Q. Ali · International Journal of Apllied Mathematics · 2025
Ensuring network security is a critical concern in today's increasingly digital world, requiring sophisticated methods to identify malicious activities. This research employs machine learning methodologies, specifically ‘Random Forest, Decision Tree, Naïve Bayes, and AdaBoost’ algorithms, to conduct a comprehensive behavioral evaluation of malicious activities in network traffic. The study underscores the critical importance of data sampling in datasets for augmenting the effectiveness of intrusion detection systems (IDS). The dataset utilized for experimentation includes the CICIDS2017 and CICIoT2023 datasets, offering a varied and authentic portrayal of network traffic situations. Using Random Forest and Decision Tree algorithms helps model complex relationships in the data, while Naïve Bayes and AdaBoost algorithms add diversity to the analysis. The study examines how data sampling techniques affect the performance of algorithms, highlighting the significance of a carefully curated dataset in developing strong intrusion detection models. The experiment results reveal the pros and cons of each algorithm, providing insight into their efficacy in detecting malicious activities. This study's results contribute significantly to the advancement of behavioral analysis in intrusion detection by providing recommendations on algorithm choice and data preprocessing methods to achieve optimal performance in practical network security situations. The results highlight the important function of machine learning in strengthening network security and offer useful suggestions for creating strong IDS systems.