Soybean production value in the Rio Grande do Sul under the GAMLSS framework

Abstract

Abstract In this article, we consider the more recent soybean production data in Rio Grande do Sul (years 2017 and 2018) and obtain regression models through the generalized additive models for location, scale, and shape (GAMLSS) approach, and provide a dashboard as a visualization tool of the considered variables. Two models are applied to explain and predict the soybean production value as a function of the covariates, such as produced quantity, number of establishments, and average yield in each city of RS. Validation and cross-validation methods are considered to assess whether the predictions provided by the fitted models are reliable. The fitted model with data of 2017 provides the best predictions. The GAMLSS framework may be more accurate than linear regression to model data related to soybean production, constituting them in a reliable and useful source to auxiliary the farmers and economic sector managers in making decisions.