Virtual Booth

Floods remain one of the most frequent and destructive natural disasters in Malaysia, posing significant threats to lives, infrastructure, and economic stability. Despite advancements in flood prediction systems, many existing models rely solely on static historical data and struggle to adapt to evolving environmental conditions. To address this limitation, this research introduces the Flood Adaptive Learning Case-Based Reasoning (FAL-CBR) model. FAL-CBR is an intelligent flood forecasting approach that dynamically improves prediction accuracy through real-time, event-based feedback. FAL-CBR enhances traditional Case-Based Reasoning by incorporating an adaptive weight adjustment mechanism. The model receives live input data which arerainfall, streamflow, and water level. It then retrieves the most similar historical cases based on similarity calculation. It then compares predictions against observed outcomes and uses the feedback to update the attribute weights, refining similarity assessment over time. This continuous learning loop allows FAL-CBR to adapt to changing flood behavior and local hydrological variability. The model was tested across six monitoring stations in southern Malaysia, and the results demonstrate a substantial improvement in prediction accuracy. FAL-CBR achieved a 35% reduction in Root Mean Square Error (RMSE) compared to traditional CBR models, showcasing its ability to deliver more reliable early warnings. By enabling smarter, self-improving flood predictions, FAL-CBR supports better decision-making, enhances community preparedness and contributing to more resilient flood risk management systems.