Comparative Analysis of Feature Selection Methods and Machine Learning Algorithms in Permission based Android Malware Detection

The most anticipated cell phone working frameworks available in the market is Google android cell phone board. The open source android board raises trivial problems related to malevolent applications (Apps) and enables designers to take full preferred standpoint of the portable activity framework. On one hand, the distinction of android assimilates consideration of most engineers for building up their applications on this board. Then again, the expanded quantities of utilizations, readies an appropriate inclined for a few clients to create distinctive classes of malware and embed them in Google android advertise or other outsider markets as kindhearted applications. The issue of identifying such malware presents an elite test because of the confined assets accessible and insufficient benefits conceded to the client, yet additionally introduces extraordinary open door in the required metadata connected to every application. Consequently, in this work, android malwares are identified based on the permissions it demands from the client. A few machine learning calculations are being utilized in the discovery of android malware based on the group of permissions empowered for each application. This paper makes an endeavor to examine the execution of different attribute selection methods, like Relief Attribute Evaluator, Gain Ratio Attribute Evaluator, Correlation based Feature Subset Evaluator (CFS), Chi-Square (CH) examination and various machine learning calculations, as Naïve Bayes (NB), J48, Random Forest (RF), Support Vector Machine (SVM), Multi-Layer Perceptron based Neural Network (MLPNN), k-Nearest Neighbor (kNN) and hence an arrangement of results acquired for permission based malware recognition and categorization demonstrates that Chi-Square attribute selection technique and SVM machine learning calculation are overtaking the other feature selection and machine learning methods correspondingly.