A Neural Network Approach for Predicting Crop Import Prices: A Case Study of Qatar

Accurate crop import price forecasting is essential for ensuring food security and guiding agricultural and trade policy, especially for countries like Qatar that heavily rely on imports. This paper investigates the application of neural networks (NNs) and ensemble techniques to predict import prices of various crops using historical trade data from the United Nations Comtrade database. We develop a NN model, tailored to forecast the price of specific crop imports, and compare the performance across different configurations of hidden layers. To enhance prediction accuracy and robustness, an ensemble method, averaging the predictions of multiple NNs, is employed. The results show that while individual NNs perform well for certain crops, the ensemble consistently improves the stability and overall accuracy of predictions, particularly for crops with more complete historical data. The study highlights the potential for incorporating additional external factors, such as long-range weather forecasts and geopolitical influences, to further refine predictions. This research demonstrates the effectiveness of NN ensembles in enhancing crop price forecasting, contributing valuable insights for agricultural decision-making and trade strategies.