Auto-regressive model for the evaluation in dairy Holstein herd

Abstract

Daily, weekly, monthly, and every 20 days milk records, obtained at milking in the morning, afternoon, and days totals from 1988 to 2003 were used to estimate milk production of Holstein cows in a herd located in Agreste region on Pernambuco state. The data set had 117,196 entries, acquired from 414 lactations of 188 Holstein cows and pure source of high genetic potential for production. The statistical analysis was performed using an autoregressive model of first order. The predicted autocorrelations between control daily, weekly, monthly, and every 20 days were overestimated in any period of lactation, especially between two forthcoming productions and in the range of one to 30 days. Regardless of the control range considered, Pearson's correlations were high and significant among milk production in the morning, afternoon, and day total. Thus, among the records used, all of them can be indicated to estimate total milk production, leaving the discretion of the definition of producer milk yields in the morning, afternoon and day total. Regardless of the control range adopted, if one chooses the day's total output, she/he should be aware of the elapsed time between two milkings to compose this production, for the management of milking and the number of milking can affect the physiology of the udder and, consequently, milk production.