DYNAMIC CLASS SCHEDULING USING REINFORCEMENT LEARNING

Dynamic class scheduling is a challenging problem for educational institutions, requiring efficient allocation of courses, instructors, and resources while adapting to the preferences and availability of students and faculty. Traditional scheduling methods are often rigid and struggle to accommodate real-time changes or diverse constraints, leading to issues like overbooked classes, scheduling conflicts, and low satisfaction levels. This paper presents a novel approach to dynamic class scheduling using Reinforcement Learning (RL), which allows the scheduling system to learn from its environment and improve over time. By formulating the scheduling problem as a Markov Decision Process (MDP), the RL-based system generates adaptive schedules that respond effectively to changes such as instructor availability or shifts in student enrollment. The proposed methodology incorporates various modules, including environment simulation, action selection, reward calculation, and constraint handling, to ensure flexibility and compliance with institutional requirements. Testing results demonstrate that the RL-based scheduling system improves resource utilization, balances instructor workloads, and aligns with student preferences, creating a more efficient and adaptable scheduling process. Future enhancements include integrating advanced data sources, improving customization options, and expanding the system’s applicability across different educational settings.