eobolo/Obolo_Emmanuel_Oluwapelumi_rl_summative

# Rwanda Textbook Distribution RL System ## Project Overview This project implements a reinforcement learning system to optimize textbook distribution decisions in Rwanda's education system. The system simulates a smart decision-making agent that evaluates school conditions and determines the most effective textbook delivery actions to maximize learning outcomes while minimizing waste. ### Problem Context Based on Rwanda's Foundational Learning Strategy (FLS), many schools face challenges in textbook access due to: - Insufficient textbook-to-student ratios - Poor coordination between government grants and actual needs - Mismatched textbook quality and curriculum requirements - Inefficient delivery logistics between rural and urban schools ### Solution Approach The RL agent operates within a custom Gymnasium environment where: - **States**: School characteristics (student count, available textbooks, teacher guides, grant usage, urgency level, location type, infrastructure, quality scores, delivery history) - **Actions**: Delivery decisions (send textbooks, hold delivery, reassign batch, send guides, send limited supply, flag for follow-up) - **Rewards**: Based on improved textbook-to-student ratios, appropriate material matching, effective usage, and addressing high-urgency situations ## Installation and Setup ```bash # Clone the repository git clone cd rwanda-textbook-rl # Install dependencies pip install -r requirements.txt # Run experiments python main.py --algorithm dqn --experiment 1 python main.py --list-experiments # See all available experiments ``` ## Usage ### Running Individual Experiments ```bash # DQN experiments (1-4 available) python main.py --algorithm dqn --experiment 1 # PPO experiments (1-3 available) python main.py --algorithm ppo --experiment 2 # REINFORCE experiments (1-3 available) python main.py --algorithm reinforce --experiment 1 # Actor-Critic experiments (1-3 available) python main.py --algorithm actor_critic --experiment …