This project introduces a mathematically grounded two-stage framework for modeling forest fire behavior. Unlike conventional assumptions of exponential growth, this study reveals that:
- Fire occurrence is a binomial ignition process
- Fire spread (burned area) is overdispersed, requiring careful probabilistic modeling
The work blends machine learning with statistical theory to reflect the true generative nature of fire behavior.
⚠️ This is a pure scientific study. It proposes no interventions or simulations — only mathematically sound modeling.