Feature Selection of Android Malware Detection and Analysis

Abstract

Mobile malware performs malicious activities like stealing private information, sending message SMS, reading contacts can harm by exploiting data. Malware spreads around the world infects not only ends users applications but also large organizations service providers systems. Android malware is prominent to study the best classifiers that can detect these malwares effectively and accurately through selecting the most suitable permissionbased features as well as comprehensive comparison with detecting android malware. This study evaluates five machine learning classifiers, namely BayesNet, MultilayerPerceptron, Decision Tree, K-nearest neighbour, and RandomaForest. The evaluation was validated using malware data samples from the Android Malware Cantagio. This paper focused on evaluating the best feature selection to be employed in the best machine-learning classifier.