Flood prediction and susceptibility mapping are critical components in mitigating the impacts of flooding, one of the most devastating natural disasters worldwide. This study aims to develop and compare advanced machine learning algorithms to enhance the accuracy and reliability of flood prediction and susceptibility mapping. By leveraging state-of-the-art techniques in data science, including deep learning and ensemble methods, this research seeks to identify the most effective models for forecasting flood events and delineating high-risk areas. The research methodology involves the collection and preprocessing of extensive hydrological and meteorological data, feature selection, and the application of various machine learning algorithms such as Random Forest, Gradient Boosting, Support Vector Machines, and Neural Networks. These models are evaluated based on performance metrics including accuracy, precision, recall, F1-score, and area under the receiver operating characteristic (ROC) curve. Additionally, the study employs Geographic Information Systems (GIS) to integrate spatial data, enabling the creation of detailed susceptibility maps. Comparative analysis of the models highlights their strengths and weaknesses, offering insights into the most suitable approaches for different flood prediction scenarios. The results demonstrate that advanced machine learning algorithms significantly improve flood prediction accuracy and provide robust susceptibility maps, which are essential for effective flood risk management and mitigation strategies. This research contributes to the development of more reliable early warning systems and supports decision-makers in implementing proactive measures to protect vulnerable communities and infrastructure from flood hazards.

Flood Prediction, Machine Learning, Cascade Forest Model (CFM) and Long Short-Term Memory (LSTM) Neural Network, Geographic Information Systems.