Agent-Based Modeling and Reinforcement Learning for Optimizing Household Waste Segregation Policies in the Philippines
This study proposes the development of an integrated Agent-Based Model (ABM) and Reinforcement Learning (RL) framework to identify the optimal policy mix for maximizing segregation compliance in a Philippine LGU. The ABM will simulate household decision-making by incorporating constructs from the Theory of Planned Behavior (Attitude, Subjective Norms, Perceived Behavioral Control), socio-demographic factors (income, education), and the economic utility of LGU policies (fines and incentives).
The model will be parameterized and calibrated using empirical data from a household survey in a Northern Mindanao LGU. An RL algorithm will then be deployed to allow the LGU agent to autonomously discover the most cost-effective policy strategy—whether pure incentive, pure penalty, or a hybrid.
🧠 Expected Outcome:
A validated computational tool and a set of data-driven recommendations on the optimal fine-to-incentive ratio, providing LGUs with an evidence-based method for policy design that improves compliance while optimizing the use of public funds.
- Solid Waste Management
- Agent-Based Modeling
- Reinforcement Learning
- Theory of Planned Behavior
- Policy Optimization
- Local Governance
- Philippines
- Model household decision-making behavior toward solid waste segregation using ABM.
- Integrate RL to discover optimal LGU policy strategies.
- Evaluate the impact of varying incentive-penalty mixes on long-term compliance.
- Provide data-driven recommendations for sustainable LGU policy design.
| Component | Description |
|---|---|
| Agent-Based Model (ABM) | Simulates household agents with behavioral, demographic, and utility-driven decision processes. |
| Reinforcement Learning (RL) | Enables the LGU agent to iteratively optimize policy interventions through reward feedback. |
| Calibration Data | Derived from empirical household survey data within a Northern Mindanao LGU. |
| Policy Scenarios | Pure incentive, pure penalty, and hybrid approaches analyzed for cost-effectiveness. |
| Tool | Purpose |
|---|---|
| 🐍 Python | Primary programming language |
| 🧬 Mesa | Agent-Based Modeling framework |
| 🧠 Stable Baselines / TensorFlow / PyTorch | Reinforcement Learning |
| 📊 Pandas, NumPy, Matplotlib | Data handling and visualization |
| 🧾 Jupyter Notebook | Experimentation and result documentation |
- A functioning ABM-RL simulation environment
- Policy optimization results and visualizations
- Recommendations report for LGUs
- Open-source repository for replication and improvement
If you use or reference this work, please cite as:
Lumingkit, J. J. J., et al. (2025). Agent-Based Modeling and Reinforcement Learning for Optimizing Household Waste Segregation Policies in the Philippines. Mindanao State University - Iligan Institute of Technology.
| Name | Role |
|---|---|
| Hussam M. Bansao | Researcher / Developer |
| Jemar John J. Lumingkit | Researcher / Developer |
This project is released under the MIT License – feel free to use, modify, and distribute with attribution.
“Clean data for a cleaner world.”