Democritus University of Thrace (DUTH)
Εφαρμογές στατιστικών μοντέλων σε δεδομένα γαλακτοπαραγωγής κτηνοτροφικών εκμεταλλεύσεων
Abstract
dc:descriptionThe basic aim of this research is the investigation of the behavior of the statistical models describing/simulating the lactation curve, with a view to facilitate management decision making. Five models were used in the current study for the description of the lactation curves. The models were fitted by non-linear regression to the data. Estimates of the parameters of each of the models were obtained for each individual lactation curve. Cluster analysis was used to investigate the nature of the lactation curves in each model. Analysis of variance was performed for the comparisons of the parameter estimates by season of lambing and number of lactations. The information criteria used for the comparison of the models were MMSE, AIC, AICc, BIC. Coefficient of determination R² and the convergence percentage for the models were calculated. The existence of statistical significant correlation with the coefficient of Pearson between parameters of the models and the typical characteristics of the lactation were also studied. In our study it is apparent that the Wood (1967) model still remains a good choice with a great suppleness for describing lactation curves with different behaviour. At the second part of the current study the utilization of the statistical models through a procedure of simulation and linear regression was attempted. A simulation procedure was used for the production of the data of our study. The generation of the lactation was based on the lactation curve of Wood model. It was created a total number of 100 herds with 100 ewes each. Our data set was analyzed using a linear model with SAS statistical software and parameters were estimated. The amount of the management intervention was calculated by a dummy variable TRT from the regression analysis. The estimated values of TRT and TRT-T were investigated by an analysis of variance across herds and level of management intervention, where Τ is the level of the known input of management intervention. T-test followed in our analysis for the evaluation of management intervention TRT and for the degree of accuracy of estimates of TRT-T. The evaluation of the intended method based on the calculations of sensitivity, specificity, accuracy, positive predictive value and negative predictive value for each test day and for each level of management intervention. The results of our analysis are satisfactory. The procedure is sensitive in detecting changes in milk production of 150 g or -150 g with percentage of sensitivity of 85% and 90% respectively, while the percentages of accuracy are 83% (150 g), 85,5% (-150 g) for the first test date. The percentages of sensitivity for the second test date are 96% and 93% respectively, and the accuracy is 90,5% (150 g), 89% (-150 g). The results of the estimates of the positive and negative predictive value are also high. The information extracted is useful for the evaluation of milk yield production and the assessment of the way of management of a dairy sheep herd.
Degree
thesis:*- Grantor dc:publisher
- Democritus University of Thrace (DUTH)
- Year dc:date
- 2012
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Καραγγελή, Μαρία
Subjects
dc:subject × 16- Καμπύλη γαλακτοπαραγωγής
- Πρόβατα φυλής Χίου
- Ημερήσια απόδοση γάλακτος
- Περιβαλλοντικές επιδράσεις
- Μεταβολή της παραγωγής γάλακτος
- Προσομοίωση
- Lactation curve
- Chios sheep
- Daily milk yield
- Environmental effects
- Milk production change
- Simulation
- Γεωπονικές Επιστήμες και Κτηνιατρική
- Επιστήμη Ζωικής Παραγωγής
- Agricultural and Veterinary Sciences
- Animal and Dairy Science
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/28677