Predictive Analytics of Plots For Soybean Production

Abstract

This study is to determine the feasible productivity of a plot of land in large fields where the quality of the soil and the weather conditions fluctuate every year, hindering optimum soybean production practices. The aim is to predict 8 sceneries through the artificial neural network model and study its reliability. Then predict 7 feasible sceneries to achieve a good sowing strategy on certain plots of land and with certain types of seeds. Finally, to make a prediction using the average historical rainfall data collected during the studied months and to observe the fluctuations on the yield in accordance with previous predictions. The artificial neural network is the method used and it was provided by soft RISK Industrial 7.6 (Neural Tools). The result is going to be compared with the data collected from the company “Nueva Castilla” of Trenque Lauquen (Buenos Aires province, Argentina) to determine the practical and technical feasibility of the model. These data correspond to more than 17 years of climate and weather analysis, soil and soybean yield with different types of seeds.