Rainfall Prediction in Diverse Indian Regions Using Machine Learning Approaches

Agriculture has an irreplaceable important position in poor countries such as India. The availability of precipitation is fundamental for farming and has an impact on water resources, agricultural ecology, and hydrology. Today, rain forecasting is as much of a key concern as other predictions in the future. Accurate rainfall forecasting accurately enlightens people with useful knowledge, preparing for probable future occurrences that may be dangerous to their crop yields and protecting against the influence of adverse weather on them. This research aims to assess the efficiency of experimental models in predicting weather phenomena, XGBoost with extreme gradient boosting technology, Multilayer Perceptrons (MLPs), Convolution Neural Networks (CNNs), and Long Short Term Memory (LSTM) networks. The Rajasthan and Kerala region dataset was collected from the National Aeronautics and Space Administration (NASA) from 1980 to 2022. Using a monthly climate dataset from Rajasthan and Kerala states, the evaluation set for these frameworks is partitioned into 80:20 for training and testing. The root-mean-square error (RMSE) evaluation method measures how effectively a model predicts an outcome from combined mathematics and statistics. The LSTM model outperformed the other three models, and its RMSE values are as follows: 0.056 for Rajasthan; and 0.041 for Kerala. These experimental results show the LSTM model's strong predictive capabilities for rainfall.