Flood Forecasting for Himachal region using Machine Learning based classification

Abstract

Globally, floods are the most frequently occurring catastrophic disasters that cause considerable environmental and economic damage to affected areas. In the last decade, it has affected millions of lives worldwide. India has been majorly affected by devastating floods from time to time. Recently occurring floods, especially in the hilly states of India, shocked the world and forced the authorities to consider alternatives for flood prediction systems. Here, Machine Learning (ML) can provide a better alternative. The field of ML is rapidly changing the world. Today, every person uses products and services that use ML algorithms. It is a technique that allows machines to process and learn from data by finding patterns. Currently, ML models are frequently and successfully used for prediction tasks by prominent organizations worldwide. The work presented in this paper is an effort to encourage the use of ML models for flood forecasting. Predicting floods in hilly areas has been considered in this research to prove the usefulness of ML models. In particular, various supervised learning classification techniques have been used to classify daily weather and rainfall data. The aim is to find the best model to perform the classification for predicting whether there are significant chances of the occurrence of floods the next day or not, given the weather data of the previous day. The idea is to train, test, and compare the performance of popular ML models for logistic regression, support vector machines, decision trees, random forests, and artificial neural networks.