Classification of Android Malware Applications using Feature Selection and Classification Algorithms

Abstract

Smartphones have become a potential part of our lives, and this led to a continued increase in the number of smartphone users. The growing number of users attracts hackers to develop malware applications to steal the private information and causing potential financial losses. Due to the fast modifications in the technologies used by malware developers, there is an urgent need for more advanced techniques for malware detection. In this paper, we propose an approach for Android malware classification based on features selection and classification algorithms. The proposed approach uses the permissions used in the Android app as features, to differentiate between the malware apps and goodware apps. The information gain algorithm is used to select the most significant permissions, then the classification algorithms NaivBayes, Random Forest and J48 used to classify the Android apps as goodware or malware apps. The experimental results show that random forest algorithm achieved the highest precision of 0.898 with a lowest false positive rate of 0.110.