Fault Prediction of Agile-Based Software Using Attention-Based BiGRU
Shikha Dwivedi, Neeraj Kumar Goyal · 2024
Software reliability is a critical concern in various industries, where failures can lead to severe consequences. Agile development, particularly under the Agile Framework (SAFe), has gained prominence for managing large-scale projects. However, there is a limited focus on predicting software reliability within an agile context. Existing models often treat each sprint independently, overlooking cumulative effects. This research introduces a fault prediction model for SAFe-based systems, utilising a Bi-Directional Gated Recurrent Unit (BiGRU) and a self-attention mechanism. BiGRU considers past and future sprints, addressing the shortcomings of traditional models. The self-attention mechanism enhances the model's ability to capture dependencies between sprints. The proposed approach demonstrates its effectiveness through a practical example. This research offers a more accurate and comprehensive strategy for anticipating and managing software faults in agile, large-scale industrial applications.