Adsa Logo White Adsa Title White

A time-series analysis of increasing milk productivity and yearly seasonality.

M. Li

Events

06-23-2020

Join M. Li on this page for a live text chat!
6:00 PM - 8:00 PM GMT

Abstract:

T78
A time-series analysis of increasing milk productivity and yearly seasonality.
M. Li*1, V. E. Cabrera1, K. F. Reed2. 1Department of Dairy Science, University of Wisconsin-Madison Madison, WI, 2Department of Animal Science, Cornell University Ithaca, NY.

US dairy cows are increasing their milk productivity due to improved genetics and farm management. Also, milk productivity shows seasonal patterns. It is critical to understand these trends and patterns for projecting production fluctuations. The objective of this study was to accurately forecast lactation performance to be used in a whole-farm simulation model, the Ruminant Farm System model (RuFaS). We quantified the seasonal effect and the trend of improvement on milk and milk component yields over the years. As input, we used a data set containing 10 million lactations records from 6 million Holstein's during the years 2006 to 2016. Each record included complete milk, fat, and protein yield standardized yearly. Although the data included 47 states, most records were from Wisconsin (26%), Pennsylvania (14%), and New York (13%). Lactation records were 40% 1st lactation, 28% 2nd lactation, and 32% later lactations. We decomposed a time series of the data with an additive model yt = St + Tt+Rt, where yt, St, Tt, and Rt were the yield, the seasonal, trend, and error terms, respectively, at time t. We found a strong seasonality effect and increasing trend for milk (MY), fat (FY), protein (PY), and energy-corrected milk (ECM) yields for all parities across the studied years (Table 1). The seasonality pattern indications (peak and trough period) and averages for the year 2016 and their change from 2006 are shown in the table. For forecasting yields in RuFaS, we used the Holt-Winters' seasonal forecasting method, which applies a triple exponential smoothing for milk production level, trend in milk production, and season, giving higher weight to the most recent data. The accuracy of forecasting was tested by identifying patterns from 2006 to 2015 and comparing our 2016 forecast with 2016 observed data. Observed values were in the 95% confidence interval of our predictions.Table 1.

�Lactation�TraitPeakTrough2016(kg/cow per yr)2016 − 2006(kg/cow per yr)
1ECMNovMar10,956908
MYNovAug10,537728
FYOctJan39035
PYNovMar32228
2ECMMarAug12,5241,145
MYMarAug12,142993
FYMarJul44243
PYMarSep37035
LaterECMJanJul12,7381,254
MYJanJul12,4041,138
FYDecJul45145
PYDecJul37239

Keywords: RuFaS, calving season, time series.