Predição de peso de bovinos Curraleiro-Pé-Duro utilizando algoritmo de arvore de decisão

Abstract

This study aimed to analyze two approaches for predicting the body weight of cattle by comparing Linear Regression and Decision Tree models. The objective was to evaluate the effectiveness of these models through both general and categorical analyses, considering morphometric variables. Four approaches with different measurements were used: thoracic perimeter, withers height, body condition score, and body length. The data were divided into 70% for training and 30% for testing. The accuracy of the models was measured using RMSE (Root Mean Square Error). The results showed that although Linear Regression presented a lower RMSE in some categories, the differences compared to the Decision Tree were not statistically significant. It is concluded that both models perform similarly, and using complete data with a smaller sample size improves accuracy, while incomplete data across a larger number of samples compromises predictive precision.