Evaluation of Machine Learning Methods for Android Malware Detection using Static Features

Popularity of android platform has made it a prime target for security threats. Third party app stores are getting flooded with malware apps. An effective way of detecting and therefore preventing the spread of malware is deemed necessary. In this paper we apply and evaluate machine learning approaches using static features to detect presence of malware in Android OS. We applied correlation based feature selection techniques and trained each classifier on the train set by hyperparameter tuning with stratified 10-fold cross validation and evaluated their performance on the unseen test set. Our experimental results reveal that it is possible to detect android malware with high reliability.