Deep Learning for Predicting Food Commodity Prices in Traditional Markets

Banyumas is one region with a history of fluctuating food prices, especially the price of shallots and chicken meat. Several factors affect fluctuating food prices, such as harvests, seasons, market demand, and so on. Farmers or ranchers often experience losses in selling crops and livestock products. This phenomenon occurs due to intermediaries monopolizing prices, and farmers need more knowledge about market prices, so traditional transactions are detrimental. Price forecasting can be one of the practical solutions to help farmers and ranchers understand the prices of their crops and livestock yields. Therefore, Deep Learning is applied to the experience of food commodities with LSTM and GRU. In this study, the LSTM model achieved the highest performance by producing an MSE value of 0.00035, while the GRU model produced an MSE value of 0.00036. LSTM models perform better with layer LSTM, epoch 100, batch size 64, optimization Adam, and neuron 128.