Abstract
1. Two different methods, categorised as input–output and single output models, were evaluated for slaughter weight prediction of broiler chickens. The input–output models included linear and non-linear recursive modelling with a time-varying model structure, whereas the output models consisted only of empirical growth equations and several growth curve fitting techniques.
2. The results suggested that a simple linear growth curve fitting method gives the greatest accuracy in a prediction horizon of 4 d or less. Error is minimised to an average of 0·14% when 4 d of past information is used to fit a line to predict the end weight one day ahead.