Generalized Unsupervised Data-Driven Modeling for Automatic Flood Prediction

Abstract

This work is aimed at a study of some parameters through the use of AI methods, with the aim of identifying dynamics that allow establishing a level of flooding risk. The choice of parameters is the result of observations obtained in India and Bangladesh which are notoriously the states with the greatest number of floods and the greatest number of losses of human lives. After processing the data for better reading and interpretation, we applied the following Machine Learning models: Logistic Regression (LR), k-nearest neighbors (KNN), and eXtreme Gradient Boosting (XGBoost). CatBoost, LightGBM. We subsequently evaluated their ability and accuracy to predict flood events. Overall The algorithms performed well in prediction.