Forecasting Severe Weather Impacts on Agricultural Supply Chains Using Multi-Variate Time Series Machine Learning Models

In this paper, we explore applying multivariate time series models using Python to forecast the impacts of severe weather on agricultural supply chains. Using data on key environmental variables from India, such as humidity, temperature, wind speed, precipitation, and severe risk, we evaluate the performance of various models, including SARIMA, Recurrent Neural Networks (RNN), Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and Decision Tree. Our results show that GRU outperforms other models where R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> is 0.917, RMSE is 6.860 and predicts weather-related disruptions over a 4-month horizon, demonstrating the ability to capture complex relationships between environmental factors. The findings emphasize the importance of incorporating multivariate analysis to support agricultural planning and mitigate the risks posed by severe weather events. Future work aims to expand the feature set and explore hybrid modeling techniques to enhance forecast accuracy.