Predicting Wildfires Triggered by Human-caused Ignition: A Spatial Framework Integrating AI Models
Аннотация
This repository accompanies a manuscript under consideration in Ecological Informatics. This provides the full dataset and source code supporting the study: “Predicting Wildfires Triggered by Human-caused Ignition: A Spatial Framework Integrating AI Models.” The materials enable full reproducibility of the analyses, including model training, validation, uncertainty assessment, and spatial risk mapping. Contents include: -National wildfire ignition inventory (2001–2025)-Preprocessed environmental, climatic, land-use, and socio-demographic predictors-Python scripts for Random Forest (RF), Generalized Additive Models (GAM), and Geographically Weighted Regression (GWR)-Model validation outputs, bootstrap-based uncertainty estimates, and feature importance results-High-resolution wildfire risk maps and high-risk area delineations All data and code are released for research and educational purposes to support transparency, reproducibility, and further development of human-caused wildfire risk modeling frameworks.