Advanced Predictive Analytics for Agricultural Weather Forecasting Using Machine Learning

Agriculture faces growing challenges due to the impacts of climate change, which affect crop yields and farm management practices. Accurate weather forecasting has become essential for farmers to mitigate risks and optimize agricultural decisions. This study proposes the use of a Long Short-Term Memory (LSTM) network to predict weather conditions based on time-series data from BMKG, focusing on parameters such as temperature, humidity, rainfall, and wind. The research involves optimizing key hyperparameters, including the number of hidden layers, neurons per layer, learning rate, batch size, and epochs, to enhance the model's predictive accuracy. After systematic tuning, the model was able to achieve Mean Squared Error (MSE) values in the range of 0–1 for most weather parameters. However, parameters related to rainfall and wind direction resulted in higher MSE values, indicating the need for further refinement. These findings suggest that future work should explore integrating additional data sources and hybrid models to improve predictions. The LSTM model developed in this study provides farmers with actionable insights, enabling them to make more informed decisions regarding planting, irrigation, and harvesting. The integration of machine learning into agricultural practices offers a significant advancement in precision farming, contributing to sustainable agricultural development and food security. By leveraging advanced predictive analytics, this study underscores the potential of data-driven approaches in transforming traditional farming practices to better adapt to the challenges posed by climate change.