Modelling and forecasting dairy milk production: Evidence from Autoregressive Moving Average (ARMA) models

Abstract

Dairy sector is one of the fastest growing sectors in the world with little global contributions from African countries and from Nigeria in particular. This study modelled and forecast diary milk production in Iwo and its environs using different variants of Autoregressive Moving Average (ARMA) models. Data used in this study comprised the daily milk production in Iwo and environs between 26th May, 2021 and 31st May, 2022 as obtained from Bowen University milk collection centre in Iwo, Nigeria. Sstationary was examined using the ADF test and p-value of. 0020 indicating the data is stationary at level (p<.05). Preliminary result indicates that milk production in the study area was as it peak in June, 2021 with 295 litres and the least value was obtained in September, 2021 with average of 63.95 litres in 2021 and 64.93 litres in 2022 respectively. There was no significant difference in litres of milk produced between these two years (p>.05). Four ARMA models, ARMA(1,1), ARMA(1,2), ARMA(2,1) and ARMA(2,2) were identified as tentative models with BIC values of 7.528, 7.550, 7.549 and 7.570 respectively. The four tentative models satisfied model diagnostic checking (p>0.05) and ARMA (1,1) gave the lowest value of Root Mean Square Error. Hence, ARMA(1,1) was adjudged as the best forecasting model among the competing models. Result of 30 days forecast shows that there will be a consistent steady increase in milk production within a range of 38 litres to 55 litres per day.