<![CDATA[Malware attacks are a significant cyber threat requiring sophisticated detection methods. This research aims to compare the performance of four machine learning algorithms (Gradient Boosting, Neural Network, k-Nearest Neighbors, and Support Vector Machine) in detecting malware on the ClaMP dataset. The research method includes data preprocessing, feature selection using Information Gain to select the Top 40 best features, and model testing across four feature subset scenarios (ALL, DDoS, File Features, and Optional Features). Model evaluation used the 10-Folds Cross-validation method to ensure result stability. The results show that the Gradient Boosting model combined with all 40 best features (ALL scenario) achieved the highest performance with 98.7% accuracy. This result surpasses the primary reference study, which recorded 96.91% accuracy using 68 features. The Neural Network model also showed excellence in the File Features scenario (9 features) with 91.9% accuracy. In conclusion, applying Information Gain feature selection combined with the Gradient Boosting algorithm proved to significantly increase detection accuracy while simultaneously reducing the number of features required (efficiency). Keyword: Malware, Machine Learning, Gradient Boosting, Feature Selection, Classification.]]>
Copyrights © 2026
