Rarh Rainfall Analytics: Machine Learning Insights from NASA Data

Predicting rainfall is an essential part of meteorology and is crucial for solving problems in agriculture, water resource management, and disaster relief. In regions with significant climatic variability, such as the Rarh region of India, accurate rainfall forecasting is essential to support sustainable development and ensure agricultural productivity. This research focuses on predicting rainfall over the Rarh region of India, an area known for its diverse climatic conditions and agricultural reliance. By leveraging NASA’s comprehensive satellite datasets, this study employs advanced ML techniques to increase the accuracy of rainfall forecasting in this region. Integrating various meteorological variables into the dataset enables the discovery of intricate patterns and relationships that traditional methods might not detect. This research adopts a systematic machine learning pipeline involving feature selection, correlation matrix analysis to understand relationships among features, 10 CV, hyperparameter passing, and three splitting ratios for training and validation. Five ML classifiers are employed to analyze performance among which Naïve Bayes gave the best accuracy score of 0.99.