Forest Fire Prediction Analysis Using the Random Forest Classifier Algorithm

Abstract

Forest fires pose significant threats to ecosystems, biodiversity, human safety, and economic sustainability, making accurate early prediction essential for effective disaster mitigation. This study proposes a forest fire prediction framework based on the Random Forest algorithm using multi-source environmental data. The proposed framework consists of data acquisition, preprocessing, feature normalization, feature selection using Random Forest feature importance, model training, and performance evaluation. Environmental variables, including air temperature, relative humidity, rainfall, wind speed, vegetation index, elevation, slope, and land cover, are utilized to represent conditions associated with forest fire occurrence. The performance of the proposed model is compared with the K-Neighbors (KNN) classifier using standard classification metrics, including accuracy, precision, recall, and F1-score. Experimental results show that the Random Forest classifier achieves an accuracy of 100%, with precision, recall, and F1-score all reaching 1.00, while the KNN classifier records an accuracy of 47% and precision, recall, and F1-score values of 0.26. These findings demonstrate that Random Forest significantly outperforms KNN by effectively capturing complex relationships among environmental variables and minimizing classification errors. The proposed framework provides a reliable and accurate solution for forest fire prediction and has strong potential to support early warning systems and decision-making in forest fire management. Future work will focus on integrating real-time environmental data, satellite observations, and hybrid explainable machine learning techniques to improve model generalization and operational deployment.