Intelligent Multi-Hazard Disaster Prediction and Alert System Using Machine Learning

Abstract

Natural disasters like floods, landslides, earthquakes and cyclones cause a lot of damage to people, buildings and the environment. So it’s really important to predict them and give people timely alerts, about floods, landslides, earthquakes and cyclones. This project proposes a machine learning-based disaster prediction and alert system that analyzes environmental parameters and historical disaster data to identify potential disaster patterns and predict their probability. Various machine learning algorithms are evaluated, among which Light Gradient Boosting Machine (LightGBM) shows improved prediction performance and computational efficiency. The system is implemented using Python and data science libraries for data preprocessing, model training, and evaluation. In addition, a hardware-based buzzer alert mechanism is integrated to notify users when the predicted disaster risk exceeds a predefined threshold. The new low cost prototype gives people warnings about dangers that’re near them. This helps people get ready for disasters and stop things from happening. The prototype is, about giving location based risk warnings so people can be better prepared for disasters and try to prevent them. The low cost prototype is a thing because it supports better disaster preparedness and mitigation.