Enhancing Android Malware Detection Through Machine Learning: Insights From Permission and Metadata Analysis

The spread of the Android ecosystem has increased potentially harmful applications, which present a serious cybersecurity threat. According to reports, a significant percentage of applications housed on the Play Store (10–24%) may behave maliciously, hiding their evil intentions behind a facade of legitimacy. The present methods for detecting malware, which are frequently resource-intensive and slow to adjust to new threats, have limits when it comes to solving this growing problem. By carefully examining data on verified dangerous applications and developing a predictive model that can reliably identify malicious applications based on the permissions they request, this study seeks to address these issues. The paper also presents a novel method for creating a machine learning-based detection model using publicly accessible metadata. With an emphasis on Android malware detection, the main goal of this study is to create a reliable model that can classify applications as malicious (1) or benign (0) depending on the permissions they request. The main activities involve a thorough examination of Android permissions and metadata as critical markers for detecting any malware, as well as the development and evaluation of a machine-learning model using publicly available metadata. In the end, initiative aims to strategically apply machine learning techniques to lessen the growing threat of Android malware. This research aims to offer a strong and proactive solution to preventing the spread of harmful programs through careful analysis and the creation of an efficient model.