Advanced Model for Predicting Weather Conditions for Smart Grape Cultivation

Smart agriculture, powered by data and advanced technologies, is transforming traditional farming practices. Given the sensitivity of grapevine cultivation to climate variability, accurate weather prediction is essential for optimizing yield and quality. This study introduces a predictive model designed to enhance smart grape cultivation through a comparative analysis between Kosovo and Iowa. The model forecasts weather conditions and determines optimal timing for grape spraying, using historical weather data and advanced forecasting techniques. Four algorithms—NeuralProphet, SARIMA (Seasonal AutoRegressive Integrated Moving Average), Random Forest Regression, and a Keras-based Artificial Neural Network (ANN)—are evaluated. The accuracy and performance of the model are evaluated using metrics like mean absolute error (MAE), root mean square error (RMSE), and mean squared error (MSE). By providing timely, data-driven insights for protective treatments, the study aims to improve cultivation efficiency, maximize yield quality, and minimize losses. The comparative approach also highlights regional climatic differences, offering tailored strategies for effective grape management.