Improving Feedlot Profitability Using Operational Data in Mortality Prediction Modeling

Abstract

Feedlot managers make difficult culling decisions using their best subjective judgment together with advice from animal health professionals. Using routinely collected operational feedlot data and five well-known classification methods, we construct mortality predictive models to aid managers in making objective culling decisions. Simulation results suggest that net return per head for calves having been treated at least once for any health incident would increase on average by $14.01 if the best-performing model were used as a culling decision aid. The probability of a positive return is 60.9%. Using cost-sensitive learning, the average value may increase to $45.27/head.