<![CDATA[Potato production in Sumatra plays an important role in supporting the national agribusiness sector. However, the uncertainty of factors such as weather, soil quality and farming practices often leads to fluctuations in production yields. This research aims to improve the accuracy of potato production prediction by applying the Backpropagation method in Artificial Neural Network. This research aims to find the best architecture model using the backpropagation method in predicting potato production in Sumatra. The data used was obtained from the Central Bureau of Statistics (BPS) and divided into training and testing data. This study tested several artificial neural network (ANN) architectures with models 6-10-1, 6-30-1, 6-50-1, 6-70-1, and 6-100-1. The results showed that the 6-50-1 architecture model gave the best results with 83% accuracy and 1617 epochs. In addition, this model has a low Mean Squared Error (MSE) on both training and testing data, which indicates good performance in potato production prediction. By using the backpropagation algorithm, this study provides an effective solution to improve the prediction of agricultural commodity production, particularly potatoes, which is important for decision-making and distribution planning. These findings can assist farmers and policy makers in planning more efficient production strategies and anticipating production challenges. This research confirms the importance of applying artificial intelligence technology to support the sustainability of the agricultural sector in Sumatra.]]>
