A Comparison of Supervised Learning Models on Distinct Datasets for Malware Detection

Malware detection remains a critical challenge as cybersecurity threats continue to evolve, particularly with the rise of memory-resident and mobile malware. In this study, we evaluate the performance of three supervised learning models-Logistic Regression, Random Forest, and Decision Tree-across two distinct datasets: CIC-MaIMem-2022 (focused on memory-resident malware) and CIC-MaIDroid-2020 (targeting Android malware). Our findings show that while all models perform exceptionally well on memory-based malware, Random Forest achieves the highest accuracy for mobile malware classification, highlighting its ability to handle complex patterns. Additionally, hyperparameter tuning further improves model robustness, demonstrating the impact of optimization in challenging detection scenarios. This study provides insights into the strengths and limitations of different models, contributing to more effective machine learning-based malware detection strategies.