As the number of Android-based devices grows in popularity throughout the world, so do thevulnerable to various attacks. The base installation package of Android is Android Package (APK). Lots of vulnerabilities are found in APK files in terms of malware and other assaults. The main focus is on predicting the malware in android APK files in order to better predict assaults with assorted vulnerabilities. To accomplish multi-dimensional outcomes, the research work incorporates the use of numerous methodologies in machine learning. With the growing number of smartphone apps and the widespread use of Android by mobile users, security concerns are becoming increasingly essential. These concerns must be addressed so that vulnerabilities can be avoided and recognised in advance. Users of smartphone apps can be linked to a warning about specific vulnerabilities using this method. Users of mobile devices can quickly install APK files from many sources without experiencing any negative consequences. A mechanism and algorithm for predicting harmful code in Android APKs must be developed and implemented. The work integrates the Android APK datasets for analytics using ensemble learning. There are two types of APKs: benign and malignant. The goal of this study is to extract deep signatures from these APKs so that a training dataset may be created. A number of APK files are analysed, with benign and the other being malignant. Then there's the checking of permissions in each APK and their consequences. Following that, the production of a cleaned dataset is planned in order to train the model for prediction. Then, in order to predict, any other random APK that isn't present in those APKs is added to the predictive analytics. Then there's a prognosis of the likelihood of having harmful key points in the new APK under investigation. The entire prediction and performance metrics are evaluated using machine learning based methodologies on a variety of parameters such as time, cost, accuracy

MachineLearning,Random Forest Approach, Support Vector Classifier, Regression and Ensemble Learning Based approaches.