Forecasting Meteorological Drought Using Weather Data With a Machine Learning Framework

Natural calamities like drought are severely impacted by the unpredictability of climate change and its adverse consequences. Drought is one of the major calamities that impact the social, economic, and environmental spheres across the globe. This necessitates accurate prediction methods to mitigate their effects. In order to forecast drought using meteorological data, this paper investigates a variety of machine learning methods, ranging from simple to intricate algorithms. The study focused on exploring the capabilities of feature selection and data balancing techniques to attain better scores. Among the different models, Random forest when combined with SMOTE achieved an accuracy of 0.82,while stacking approach emerged as the most effective with an accuracy of 0.83 and F1-score of 0.71. These findings underscores the importance of machine learning frameworks to tackle the challenges posed by drought and take proactive methodologies to mitigate its adverse impacts.