Deteksi Malware pada Perangkat Android Menggunakan Ensemble Learning

The increasing use of permission-based applications on mobile platforms has raised concerns regarding privacy and security. Android, being one of the most widely used operating systems for interacting with mobile applications, is particularly susceptible to various security risks that must be promptly addressed. Low digital literacy and a lack of user awareness about security risks—especially when installing applications from unofficial sources or without paying attention to access permissions—make users vulnerable to malware attacks. Uninformed users can easily become victims of malware insertion by irresponsible parties, turning them into targets for data manipulation and even data theft, which may then be sold on illegal forums. Attackers exploit the permission system, allowing them to freely access the target smartphone. This lack of awareness among users increases their vulnerability to malware injection and subsequent threats such as data manipulation and the theft of personal information, which can be traded on underground markets. One approach to detecting malicious behavior in mobile applications is the use of machine learning techniques. These techniques can analyze application patterns and behaviors based on features such as requested permissions. Popular algorithms for malware detection include Support Vector Machine (SVM) and Random Forest (RF), both of which have demonstrated strong performance in various studies. However, to further improve accuracy and reduce classification errors, ensemble learning approaches such as Adaptive Boosting (AdaBoost) are increasingly being adopted. Ensemble learning combines multiple predictive models to produce more reliable classification results compared to single models. This study evaluates the performance of several classification algorithms in detecting malicious Android applications. The results show that AdaBoost achieved a high accuracy rate of 91.65% and an AUC value of 95%, effectively distinguishing between safe applications and malware. Therefore, the use of machine learning algorithms—particularly ensemble methods like AdaBoost—can serve as a promising solution to enhance the security and privacy of Android-based mobile application users.