Coffee is a prominent commodity in Indonesia, serving as a raw material for the production of food, beverages, and various other products. The unpredictable nature of its production significantly impacts production costs and the economy of farmers. Therefore, it is crucial to make predictions to determine the likelihood of the development of coffee productivity in the future. This research utilizes coffee production data from farmers spanning the years 2014 to 2023. The algorithm employed is a backpropagation neural network with a 4-2-1 network architecture model and a learning rate in the range of 0.1-0.9. The training process resulted in optimal weights, with a Mean Squared Error (MSE) of 0.007 at a learning rate of 0.2. The findings of this research, it is expected to provide benefits to farmers as a solution, reference, and evaluation material to enhance coffee productivity, minimize production costs for land and plant processing, and offer information regarding upcoming harvest results.
