Use of ARIMA models for predicting milk production. Case study in UBPC ¨Maniabo¨, Las Tunas

This study was carried out to predict milk production, in a short term, by using ARIMA models, in a case study performed at the Basic Unit of Cooperative Production (UBPC, initials in Spanish) “Maniabo”, Las Tunas, Cuba. For that purpose, data of monthly milk production during the period of 2000-2010 were used. Charts of horizontal plotting and Box–Plot were used for the descriptive analysis of the series and for detecting the tendency to polynomial softening. Using periodgram, the marked peaks in the periodical movements of the series and the significance of a frequency were confirmed, and seasonal nature was detected, which was described with seasonal indexes. Correlations with the differentiated series were determined and parameters p and q of model were estimated. The best fitted model was ARIMA (1, 0, 3) x (0, 1, 0) 12 with constant. Milk production for 2011 was predicted and validated. Key words: time series, ARIMA models, milk production