Prediction of 305 days lactation milk yield from fortnightly test milk yields in hill cattle under ield conditions

Abstract

The present investigation was undertaken to predict 305 days lactation milk yield from fortnightly test day milk yields using regression and principal component analysis approaches. Accuracy of different methods i.e. simple and multiple regression and principal component analysis using simple and cumulative fortnightly test milk yields was accessed by predicted value, standard error of predicted value, adjusted predicted value, studentized residual, deleted residual, studentized deleted residual, mahaboliny distance, cook's distance and centered leverage values. Test day milk yield were recorded on 688 cows during September, 2009 to June 2010 from 27 villages of Ukhimath block (Kedarnath valley), Rudraprayag district, Uttarakhand, India. Test day milk yields were recorded by fortnightly after 5- 10 days of calving. Daily milk yield (Test day) were computed as sum of morning and evening milk yield on test day. A total of 200 cows completed 20 test day recordings under the study. The 305 days lactation milk yield was estimated from the selected data set (200 cows) those had completed 20 test day yields records using Test Interval Method. Prediction equations were developed considering 305 days lactation milk yield as dependent variable and first 12 fortnightly test day yields as independent variables. All the analysis was carried out by SPSS and SAS programs. It was concluded that for prediction of 305 days lactation milk yield, principal components analysis method using fortnightly test day milk yields was best.