Moustafa F. Ali, Angie M. Ceniza-Canillo
Software development methodology selection remains a critical challenge in project management, particularly in dynamic environments where stakeholder preferences conflict and project conditions evolve. Existing approaches rely on static decision frameworks that fail to capture the temporal dynamics and multi-stakeholder nature of methodology selection. This paper introduces MARS (Multi-Agent Reinforcement learning for Software methodology Selection), a novel framework that models methodology selection as a multi-agent reinforcement learning problem where autonomous agents represent different stakeholders (developers, clients, managers, QA teams) and learn to negotiate optimal methodology choices through iterative interaction. The framework employs Deep Q-Networks (DQN) with prioritized experience replay for individual agent learning, combined with a consensus mechanism based on Nash equilibrium for multi-agent coordination. Experimental validation across 100 simulated projects with varying stakeholder configurations demonstrates superior performance compared to traditional approaches. MARS achieves 87.3% methodology selection accuracy, reduces consensus time by 64%, and maintains 92.1% stakeholder satisfaction across diverse project scenarios. The framework adapts to changing project conditions in real-time, with agents learning optimal negotiation strategies that balance individual preferences with collective project success. Real-world deployment in three software companies shows measurable improvements: project delivery time reduced by 18%, team satisfaction increased by 31%, and methodology switch frequency decreased by 52%. These results establish multi-agent reinforcement learning as a viable paradigm for dynamic, stakeholder-aware software methodology selection that evolves with organizational learning and changing project landscapes. © 2025 Copyright held by the owner/author(s).
Department of Computer, Information Sciences and Mathematics, University of San Carlos, Cebu, Philippines
