Forecast of the gross domestic product of the state of Mato Grosso as a function of soybean, corn, cottonseed, and beef arroba production with the use of artificial neural networks

Abstract

The GDP forecast facilitates the decision-making process by relating the data obtained, such as the allocation of MT financial resources in areas of development and taxation, for example. The aim is to answer the question of the predictive power of Artificial Neural Networks (ANN) in explaining the GDP of MT about the independent variables, such as soybean production, corn production, cotton seed production, and cattle slaughter. Thus, the general objective is to estimate the GDP relative to the production of these four variables using ANNs. With the specific objective of observing the evolved correlations between these variables in relation to the production in MT; to build, train, and validate an ANN model with these variables; to use the built and validated model to make future predictions. The relevance lies in the use of this prediction tool to forecast the monetary flow that will be demanded with the increase in the production of these products, allowing for short and long-term investment planning. The article is classified as exploratory descriptive, and the time series under analysis comprises the years from 2000 to 2022. The software used for building the ANN is SPSS - Statistical Package for the Social Sciences. The predictive ANN model is capable of explaining the behavior of MT's GDP based on the described independent variables, furthermore, the model provides satisfactory values and observations for its validation. The proposed and validated models are used to make predictions with significant confidence using hypothetical data for the independent variables.