Componentes de efeitos de safras representados em biplots corrigidos por predições de modelos GEE na classificação granulométrica de cafés

Abstract

In a granulometric analysis of coffee beans with different defect categories, the data can be organized in contingency tables and, considering discrimination by crops, they might present a structure that suggests a more complex model when it comes to the interaction of crop effects with the defect classifications and percentage of sieve beans. In view of the foregoing, the hypothesis that correlation structures may be incorporated to a model in order to improve multidimensional graphic analysis (such as the biplots technique) arises. Therefore, this work has as its objective to propose the use of biplots corrected by predictions of GEE models in the granulometric classification of coffee beans, discriminated by components of crop effects. To validate the proposal, Monte Carlo realizations were performed in different contingency table structures in scenarios with different degrees of correlation. It was concluded that the use of GEE models with the biplot technique corrected by the predictions is applicable in the granulometric analysis of defective coffee beans, with an efficient discrimination of crop effects.