Predição de Indicadores Zootécnicos de Carcaças Bovinas a Partir de Variáveis de Cria

This paper describes a method for obtaining decision trees for predicting carcasse zootechnical quality indicators for bovine based on their breeding data. For such, data mining classification tasks were performed after data preprocessing, where all numeric attributes were discretized by non-equal frequency binning or by cluster discovery in distinct classification experiments. Obtained results showed that clustering techniques as means for discretization may generate classes in better balancing conjecture when in comparison to the non-equal frequency binning method, allowing the discovery of models that may be applied to real world problems.