Machine Learning Finds Sustainable Doughnut Policies
Stefano Vrizzi submits a proof-of-concept on 1 December 2025 showing how machine learning can be applied to a simple macroeconomic 'Doughnut' model to identify policy parameters consistent with living within social and planetary boundaries. The paper tests frugal methods including a Random Forest classifier and Q-learning, demonstrating these methods can find combinations that meet sustainability targets and that reinforcement learning can map optimal trajectories in parameter space. Authors propose applying methods to more complex ecological macroeconomic models.
Key Points
- 1Demonstrates Random Forest and Q-learning find policy parameters that meet Doughnut social and planetary boundaries.
- 2Highlights significance: frugal ML methods can efficiently explore macroeconomic parameter spaces for sustainability outcomes.
- 3Suggests practitioners can use RL to identify optimal policy trajectories toward Doughnut-compliant economies.
Scoring Rationale
Shows practical ML approaches for Doughnut modeling, but remains a limited proof-of-concept from a single preprint submission.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
