Abstract
Android applications have become increasingly pop-ular, with billions of users relying on them for various tasks. However, the presence of bugs in these apps can lead to poor user experience, security problems, and financial losses. Bug prediction techniques offer a proactive approach that identifies potential bugs before they manifest in released applications. This paper presents an approach for bug prediction in Android apps by leveraging historical measures and applying the random forest algorithm. Through our evaluation of twelve real-world Android applications, we have demonstrated that our approach can effectively predict buggy files. This indicates the importance of historical data when assessing Android applications' quality and reliability. In addition, we have explored the importance of each historical measure in bug prediction. The corresponding results can provide insights guiding developers in selecting the important measures to build effective bug prediction models for Android applications.