Method of predicting fire water demand based on big data

This study proposes a big data–driven method for predicting urban fire water demand using Random Forest Regression. Multi-dimensional data—including temperature, wind speed, building type, population density, and historical fire records—were collected, cleaned, normalized, and integrated to build the model dataset. The Random Forest framework effectively captured nonlinear interactions among variables, with building type and wind speed identified as the most influential factors. After hyperparameter optimization, the model achieved strong predictive performance, with a mean squared error of 537.8 liters2, mean absolute error of 22.6 liters, and mean absolute percentage error of 2.05%. The coefficient of determination (R2 = 0.865) confirmed high explanatory power. Cross-validation further verified the model’s generalization ability and robustness under diverse fire scenarios. The results demonstrate that the proposed model provides a reliable and data-driven solution for forecasting firefighting water demand, supporting resource allocation and emergency planning. This research offers valuable guidance for integrating big data analytics into intelligent fire protection and urban safety management.