Modelling the drinking patterns of young pigs using a state space model

In normal situations pigs show a stable diurnal drinking pattern. Based on experimental data, a dynamic model is developed for prediction of the drinking behaviour of growing pigs. A state space model with cyclic components is proposed for modelling the diurnal drinking pattern, measured as hourly sums. Determination of variance structure by use of discount factors is suggested. Model performance is investigated by error analysis. The final model contains three cyclic components.