Integrating Weighted Averaging and Boosting in Ensemble Models for Improved Flood Forecasting

Precise flood predictions are crucial in safeguarding the environment through effective water resource management and readiness for disasters. This research examines how combining weighted averaging and boosting methods in ensemble models can improve the accuracy of flood predictions. We assess the efficiency of various machine learning algorithms like Logistic Regression, Multilayer Perceptron, Stacked Support Vector Classifier, K-Nearest Neighbours, Random Forest, XGBoost, AdaBoost, and Hybrid Ensemble model by analyzing multiple metrics such as accuracy, recall, F1-score, ROC score, Cohen’s Kappa, and MCC. The hybrid Ensemble model stands out as the best performer, producing impressive results in all areas, highlighting the effectiveness of ensemble methods in predicting floods. Though there have been improvements, obstacles remain in the areas of data precision, computational speed, and understanding model outputs. It is important to tackle these challenges in order to improve the reliability and use-fulness of flood prediction models. By utilizing ensemble methods and enhancing machine learning approaches, this study adds to the continuous work in enhancing flood forecasting precision for efficient environmental preservation and sustainable growth.