Abstract
As mobile device usage grows, it is worth noting that smartphones are among the most important inventions of the century. The evolution of smartphones and access to affordable internet has made technology an integral part of our daily lives. Android operating systems have provided an adaptable environment for hackers to develop new mobile applications loaded with malware through which attacks such as denial of service and privacy breaches are executed. Malware developers exploit vulnerabilities in the installation and runtime files to execute cyberattacks on the devices. The present study adopts a graph-based machine learning algorithm to manage imperative permissions and API functionalities using application data from the Drebin project, in which 15,036 applications were tested to determine the most important features for malware detection. Machine learning techniques such as Logistic Regression Algorithm (LR), Decision Tree Algorithm (DT), K-Nearest Neighbor Algorithm (KNN), and Random Forest (RF) Algorithm are used in the classification and training of malware detection programs. The findings suggest that the RF technique achieves the highest rate of recall (96%) and accuracy (97%) while KNN and DT deliver (96%) accuracy while LR delivers (95%).