Application of the hierarchic markovian decision processes in the decision making processes of pig keeping

Abstract

In this study we discuss the Markovian chain-based decision processes and their developed variant called Hierarchic Markovian Processes. The optimizing possibilities of such processes are presented in detail. Moreover, we introduce a free available software based on these processes and developed by Danish researchers for supporting decisions in animal breeding. Among the several models the reduced sow model (with gestation) were chosen for presentation. We describe the basic settings and parameters for running the software as well as we calculate the average net return over time and the series of decisions per sow in case of simulated sow herd data by applying the value iteration technique. We also present the results of decisions on keeping an animal in production as well as on determining the number of matings of a sow. We also give examples of the development of the relative utility values related to such decisions.