With the widespread adoption of Android devices, concerns about security have also escalated due to the growing number of malware threats. Traditional detection methods, based on static and dynamic analysis, have shown limitations in fully addressing these issues. In such a situation hybrid approaches offer a promising solution by mitigating the weaknesses of both the individual techniques. Additionally, repeated usage of certain features across malware and benign classes underscores the importance of identifying and ranking features that effectively detect malware. Hence, in this paper, we introduce a novel Android malware detection framework, “AndroV-Rank”, which leverages permissions and system calls to identify the best set of class-distinguishing attributes achieving higher detection accuracy. These features are ranked using the VIKOR method, a Multi-Criteria Decision-Making approach based on frequency, attributes and criteria. A hybrid detection algorithm, combining machine learning and deep learning techniques, is utilized further to optimize feature selection and enhance detection performance. Our experimental results demonstrate that the proposed model achieves an accuracy of 96.55% with just 65 features, reducing the feature set to approximately 24.9% of the total. This significantly surpasses the performance of static or dynamic analysis methods when used in isolation. Our findings highlight the effectiveness of the hybrid model in improving malware detection and addressing the growing threat to mobile cybersecurity.