Modeling and predicting water consumption in fattening pigs using autoregressive moving average with external parameters

Abstract

The effective monitoring of water consumption patterns is crucial to enhance health and welfare in pig production. Understanding its dynamics and influencing factors may support early detection of welfare issues and optimize farm management. This study analysed historical water consumption data in fattening pigs and developed a predictive model to capture these dynamics. A retrospective study was conducted using thirty time series of hourly water consumption records collected over a three-year period from five batches of fattening pigs. An autoregressive moving average model with exogenous variables (ARMAX) was used to account for the autocorrelated structure of the data and to include environmental and physiological covariates such as temperature, ammonia concentration and fattening day. Water usage progressively increased throughout the fattening period, following a circadian rhythm. External factors, particularly ambient temperature, significantly affected both long-term trends and daily fluctuations. Two distinct consumption patterns were observed depending on farm temperature conditions. The ARMAX model demonstrated strong accuracy in predicting water usage, effectively capturing trends and deviations in water use. Water consumption in fattening pigs is shaped by internal time patterns and environmental factors. Using an ARMAX model, we integrated autoregressive and moving average components with key covariates to accurately predict hourly usage. This approach demonstrates the value of combining diverse data streams to enhance animal health and welfare management.