Gradient boosting with extreme-value theory for wildfire prediction

2021/10/18 by Jonathan Koh, Koh, Jonathan · 4 citations
Environmental Science · #Applications (stat.AP) #FOS: Computer and information sciences #Fire effects on ecosystems #Landslides and related hazards

paper · pdf · doi:10.48550/arxiv.2110.09497

openalex publication_date 2021/10/18 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

This paper details the approach of the team Kohrrelation in the 2021 Extreme Value Analysis data challenge, dealing with the prediction of wildfire counts and sizes over the contiguous US. Our approach uses ideas from extreme-value theory in a machine learning context with theoretically justified loss functions for gradient boosting. We devise a spatial cross-validation scheme and show that in our setting it provides a better proxy for test set performance than naive cross-validation. The predictions are benchmarked against boosting approaches with different loss functions, and perform competitively in terms of the score criterion, finally placing second in the competition ranking.

Cited by

• XGBoost meets INLA: a two-stage spatio-temporal forecasting of wildfires in Portugal
• A simulation study comparing statistical approaches for estimating extreme quantile regression with an application to forecasting of fire risk
• Predicting hazards of climate extremes: a statistical perspective
• Extreme Conformal Prediction: Reliable Intervals for High-Impact Events

Related

• Wildfire risk forecast: An optimizable fire danger index
• Regression modelling of spatiotemporal extreme U.S. wildfires via partially-interpretable neural networks
• FireScope: Wildfire Risk Raster Prediction with a Chain-of-Thought Oracle
• Reconstruction of Incomplete Wildfire Data using Deep Generative Models
• Joint Modeling and Prediction of Massive Spatio-Temporal Wildfire Count and Burnt Area Data with the INLA-SPDE Approach