Implementing Machine Learning Techniques to Forecast Floods in Bangladesh

Flooding is a complex phenomenon that demands robust predictive techniques due to its mercurial nature. This study investigates the application of Machine Learning (ML) methodologies in enhancing flood prediction accuracy, addressing associated uncertainties and unveils their efficacy in discerning complex weather and hydrological patterns. Leveraging historical data based on numerous basins in Bangladesh, the algorithms used in our research exhibit promising capabilities in forecasting future flood occurrences. Specifically, employing the Random Forest model yields notable results, with mean accuracy, precision, and Fl-Score reaching 92.8%, 77.2%, and 70.0% respectively. It underscores the potential of ML in optimising flood preparedness and response strategies, facilitating more efficient emergency management and aid distribution to affected regions.