Because of Android’s popularity as a mobile operating system, there is a growing threat to user security and privacy in the form of Android malware. In light of the low detection effectiveness of typical machine learning approaches, this paper presents an ensemble learning-based Android malware detection strategy. We have analyzed different machine learning models and selected the six best-performing models as baseline estimators and Logistic Regression as the final estimator for our ensemble model. We have created two sets of datasets from benign and malware Application Package Kits (APKs) downloaded from Androzoo and Virushare. We employed reverse engineering and analyzed nine different constructs of APKs, such as Permission, Services, System commands, for our first dataset. We extracted Dalvik Executable files from APKs, converted them into RGB images, and employed the pre-trained VGG16 model to extract pixel-based features for our second dataset. We analyzed the performances of the individual baseline estimator and ensemble model on two datasets. The proposed architecture achieved 93.84% accuracy on the dataset created by the application of reverse engineering techniques on APKs. The same model achieved 94.82% accuracy on the dataset prepared by the employment of a pre-trained VGG16 model for pixel-based feature extraction from color images generated from APKs.