Flood prediction using ensemble machine learning models

Abstract

Frequent and devastating floods in India pose a significant threat to people and property. Accurate and real-time forecasting of floods is essential to mitigate their impact. This thesis focuses on evaluating di↵erent machine learning models for flood prediction in India. The models assessed include K-Nearest Neighbor (KNN), Support Vector Classifier (SVC), Decision Tree Classifier, Binary Logistic Regression, and Stacked Generalization (Stacking). The researchers trained and tested these models using a rainfall dataset. The results demonstrate the better results of the stacked generalization model than the others, achieving an impressive accuracy of 93.3 per cent with a standard deviation(sd) of 0.098. These findings highlight the potential of machine learning models to provide precise and timely flood predictions, empowering the local authorities, specially disaster management ones, to take necessary actions to avoid destruction and preferably save people.