Synchronization of XGBoost and Random Forest Algorithms for Optimizing River Level Prediction

Flood prediction represents a critical challenge in natural disaster management, as timely risk recognition enables better preparedness and minimizes damage. This paper focuses on the application of synchronization methods for two individual controllers that utilize XGBoost and Random Tree algorithms for flood prediction. The controllers analyze the same input parameters, such as Temperature, Humidity, Precipitation, Dew Point and River Level, to provide accurate and robust predictive analysis. The main objective of this study is to achieve effective synchronization between the two models, ensuring more consistent and precise results. To this end, a dynamic graphs, triadic formation, and community addition for synchronization, enhancing integration and information exchange between the controllers are employed. The methodology includes an analysis of historical data and real-time forecasting, allowing a balanced adjustment of the models. The application of this strategy can significantly improve the early warning system for floods, thereby increasing the effectiveness of protective measures and water resource management. The expected results indicate that the combination of XGBoost and Random Tree, along with synchronization techniques, enables more accurate predictions, which is essential for reducing flood risks.