Forecasting model for improving agricultural yield using recurrent neural networks

Abstract

The agri-food industry is currently migrating to Agri-Food 4.0. This trend involves incorporating the latest technological innovations with the goal of optimizing resources and making better decisions thanks to the information and data obtained through them. In this sense, this technology, known as Agroindustry 4.0, will allow us to develop crop yield estimates, which are vital for good agronomic management. This will allow for better land management in the face of increasing food demand due to constant population growth. This research proposes a crop yield forecasting model using the Cross Industry Standard Process for Data Mining (CRISP-DM) method with the Nonlinear Autoregressive Network with Exogenous Inputs (NARX) methodology using artificial neural networks (ANN). Applying this methodology will allow for better analysis of information on possible production scenarios for a given crop. This will help agricultural policymakers and stakeholders involved in food and nutrition security, project the real-time performance of a given crop's production yield. This proposed model promises to yield results that provide significant contributions to addressing the problem of crop production, prediction and management. Ultimately, we can affirm that this study will allow us to achieve sustainable and responsible agriculture.