Predicting Flood Incidences in Kerala Using Machine Learning Models

One of the most important and lethal natural disasters of this century is flooding. The deficiency of a comprehensive flood forecasting system has resulted in significant losses of life and property. This has highlighted the necessity of having a reliable flood prediction system. This paper examines the development of the most efficient flood determining model, which combines the use of Machine Learning technique with a robust, efficient and accurate flood expectation framework to provide essential aid and assistance to the local population and government. We have used K-Nearest Neighbors (KNN) classifier, Logistic Regression, Ensemble Learning Methods, Decision Trees, and Random Forest in order to determine flood. SAR is also used to map flood affected areas. To this end, Logistics Regression consistently provided the most accurate (87.5 % accuracy) flood detection outcomes. The framework helps the public authority set aside funds for salvage works and start moving chores before the flood causes a disturbance in town.