Artificial Intelligence and Machine Learning Approaches for Flood Prediction in Medan City

Climate change has dramatically altered weather patterns in Indonesia, resulting in more frequent and severe flooding, particularly in urban areas such as Medan City. Effective flood prediction models are crucial for improving preparedness and mitigating the risks to both infrastructure and communities. This research applies machine learning techniques to predict flood occurrences in Medan, utilizing key weather parameters such as rainfall, temperature, and humidity. The method used is the Cross-Industry Standard Process for Data Mining (CRISP-DM) methodology ensures a systematic and structured approach to predictive model development. This framework enhances the robustness and replicability of the results, ensuring that the model is not only accurate but also adaptable to varying data inputs and conditions. A machine learning classification approach is employed, with the CatBoost algorithm selected for its high performance, especially in handling categorical data. The resulting model demonstrated exceptional predictive capabilities, achieving a 96% accuracy and an F1-score of $\mathbf{9 6 \%}$ in classifying flood events. The results of this research are for the development of an interactive dashboard designed to provide flood risk information.