Main Article Content
In this paper a different method for detecting android malwares is proposed. Here we use machine learning algorithms as opposed to conventional signature based method. To improve the accuracy we use hash functions like MD5, SHA 256, SHA1 along with the supervised machine learning methodology. Three classifiers namely Random Forest, Decision Tree and Logistics Regression are used. Three classifiers are used because we developed a new algorithm with these classifiers which help to reduce the false negative rate. To train these classifiers 19 static features which are highly distinguishable between malware and benign files are used .After the model fitting the classifier the detector is tested with many tests APK files and effectiveness of the detector is measured.