| The advent of new technologies has allowed commercial dairy producers the opportunity to collect a vast amount of data on individual cows. Although herd management software has been used on farm in a historical context to view cow records, little research in real-time monitoring tools for a commercial dairy farm has been conducted.;The first objective of this thesis was to characterize the variability in data streams on dairy farms exposed to different management strategies. In the first two experiments conducted in an experimental research setting, tall fescue hay replaced corn silage or alfalfa silage in a lactating cow TMR, with effects on fiber digestibility, animal performance and animal behavior observed. Tall fescue hay incorporation improved fiber digestibility with similar cow performance. Cow behavior was similar across most parameters, including rumination, physical activity, and meal patterns, but meal frequency and DMI was decreased in cows consuming TMR's with tall fescue incorporated.;The next studies sought to characterize variability in milk production, rumination, physical activity, DMI, and bodyweight on two commercial dairies with different management styles. For both dairies and most data streams, standard deviation among cows was greater than standard deviation among days within cow, and among day within pen.;A secondary objective of this thesis was to begin to use statistical process control (SPC) techniques to determine the frequency of out-of-control data points. Using the two commercial dairies from above, univariate SPC analyses were performed to determine the frequency of out-of-control data on each farm.;The last objective was to use multivariate SPC techniques to begin to integrate data streams and create a real-time analysis on commercial dairy farms. Data from a commercial dairy was analyzed via the MVP procedures in SAS to integrate data via a Hotelling's T2 control chart.;Integration of nutritional data streams on a commercial data could improve profitability on a commercial dairy by early recognition of an out-of-control process. Statistical process control, particularly multivariate SPC techniques, could integrate data streams and create a real-time analysis of a process. More research on events triggering an SPC alert and sensitivity analyses of the control charts should be conducted to improve the procedure. |
