Prediction of Flood and Planner Towards Emergency Response

Abstract: FloodGuard is a mobile-based flood prediction and alert system designed for early disaster warning. The application uses machine learning models such as Random Forest, XGBoost, and TensorFlow Neural Network to predict flood risk using weather and environmental data. Data preprocessing techniques including SMOTE, StandardScaler, and cross-validation improve prediction accuracy. The final model is converted to TensorFlow Lite for on-device execution in a Flutter application. The system collects live weather data, analyzes flood risk and sends alerts to emergency contacts during risky situations. FloodGuard provides a fast, low-cost, and user-friendly solution for flood safety and preparedness.

Keywords: flood prediction, mobile app, machine learning, Random Forest, XGBoost, TensorFlow Lite, SMOTE, StandardScaler, Open‑Meteo, emergency alerts