Data-Driven Rainfall Prediction for Smart Agriculture: A Machine Learning Perspective

The seasonal rain in Bangladesh has been very important in supporting the agricultural sector of the country, but because of climate change, it has become more erratic. In this research, the authors used historical meteorological data, 2013 to 2022, to create and test four machine learning algorithms, AdaBoost, Random Forest, Decision Tree, and K-Nearest Neighbors (KNN) to predict rainfall with sufficient accuracy. Random Forest proved to be the best-performing model having the highest accuracy, precision, and F1-score, and AdaBoost has a high recall, meaning that it is more effective in identifying events of rainfall. The results emphasize the potential of ensemble learning methods to aid smart agriculture to enhance the irrigation scheduling, counteracting weather-related risks, and enhancing food security. The model proposed helps in developing a climate-resilient decision-support model that will help farmers and policy makers make sound decisions in agricultural planning as the rainfall patterns continue to be highly unpredictable.