A Neural Network-Based Predictive Model for Forest Fire Management in Michoacán

This study presents a predictive model based on neural networks to estimate the area affected by forest fires in Michoacán, with the aim of optimising resource management. A total of 929 historical fire records (2015–2024) from 18 localities were analysed, including six within the Monarch Butterfly Biosphere Reserve. Using variables such as operational cost, fire duration, impact, and firefighting personnel, several machine learning models were evaluated, including linear regression, decision trees, random forest, and a neural network. The neural network achieved the highest performance (R² = 0.84) and identified firefighting personnel, impact, and duration as the most influential factors. Although the dataset lacked key environmental variables, the neural network demonstrated a comparatively strong predictive capacity, suggesting potential applicability in future studies employing more comprehensive datasets. Overall, the findings provide a methodological basis that may support improvements in planning and decision-making for wildfire prevention and response when operating under real-world data constraints. Smart citations: https://scite.ai/reports/10.61467/2007.1558.2026.v17i1.1156Dimensions.Open Alex.