Proposal of an agricultural neural network model for the prediction of commodity sales prices

In this paper, we are proposing a neural network model applied to the agricultural domain for predicting the selling price of raw materials. To achieve our objective, we have reviewed the various works on cost prediction related to Depp to our knowledge. The different types of neural network were also presented. Our approach to implementing this neural network consisted in collecting, preparing the data making up the network inputs, selecting and designing our neural network model. An architecture defining the different layers of the network and a mathematical model were proposed. A pseudo code was also proposed, taking as input a dataset containing historical commodity prices, economic variables (inflation, interest rates, etc.), geopolitical variables (sanctions, conflicts, etc.) and target sales prices (to be predicted). Simulation results of our proposal on a dataset, based on the kola nut cultivation domain has shown that our model can instantly predict raw material sales prices.