No abstract

The strategic logistics of agricultural production and storage aggregates information related to production and storage. In this sense, time, location, and distance from producer and consumer markets are considered, emphasizing the importance of grain storage and production logistics. The Natural Neighbor and multiquadric equation are spatial interpolation methods used to predict these variables value at non-sampled locations, for asymmetric and categorical data, respectively. This study investigated the spatial prediction of grain production (tons) (soybean, first crop corn, second-crop corn, and wheat) in the 2016/2017 growing season and qualitative data on the static capacity of warehouses in the 2017/2018 growing season. The result obtained through the spatial interpolation using the natural neighbor method was coherent, as it showed the high variability of grain production relative to the different meso-regions. Therefore, the method was appropriate because it allowed predicting the behavior of grain production in the 2016/2017 growing season in the state of Paraná-Brazil, making it possible to identify regions of higher or lower production. The result of the spatial interpolation using the multiquadric equations allowed identifying a higher predominance of storage units with a low static capacity of warehouses, but also enabled the detection of regions with a static capacity of warehouses that varied from the medium to the high category in the state of Paraná, Brazil.

The vast relevance of applications of spatial regression models has recently captured the interest of Economics and Agriculture, in the sense of better understanding the spatial behavior of the region under study, in the different forms of approaches. It is interesting to understand why some regions show greater variability than others, and why some forms of regional development are better explained. It is up to the researcher to understand, explore, and organize a series of observations, so that it is possible to make predictions, diagnoses, and recommendations to public policy managers and regional development agents. The municipalities’ Gross Domestic Product (Gdp) has driven studies involving spatial information. The objective of this study was to analyze the Gdp of the municipalities in Paraná-Brazil, in 2018, regarding soybean yield, corn yield, pig production, and the tax on the circulation of goods, through different approaches of spatial regression models. SAR and CAR models are global models, while the GWR model is considered a local one. Three spatial analysis models were used to perform this study: Spatial Autoregressive (SAR), Conditional Autoregressive (CAR), and Geographically Weighted Regression (GWR). The results were compared using the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), Cross-Validation Criterion (CVC), and the descriptive graphic of residual diagnoses-Worm Plot. The best result obtained was for the GWR model, which best explained the GDP of the state of Paraná-Brazil in terms of its covariates.

The vast relevance of applications of spatial regression models has recently captured the interest of Economics and Agriculture, in the sense of better understanding the spatial behavior of the region under study, in the different forms of approaches. It is interesting to understand why some regions show greater variability than others, and why some forms of regional development are better explained. It is up to the researcher to understand, explore, and organize a series of observations, so that it is possible to make predictions, diagnoses, and recommendations to public policy managers and regional development agents. The municipalities’ Gross Domestic Product (Gdp) has driven studies involving spatial information. The objective of this study was to analyze the Gdp of the municipalities in Paraná-Brazil, in 2018, regarding soybean yield, corn yield, pig production, and the tax on the circulation of goods, through different approaches of spatial regression models. SAR and CAR models are global models, while the GWR model is considered a local one. Three spatial analysis models were used to perform this study: Spatial Autoregressive (SAR), Conditional Autoregressive (CAR), and Geographically Weighted Regression (GWR). The results were compared using the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), Cross-Validation Criterion (CVC), and the descriptive graphic of residual diagnoses-Worm Plot. The best result obtained was for the GWR model, which best explained the GDP of the state of Paraná-Brazil in terms of its covariates.