The use of partial least squares regression procedure to determine the relative importance of the lifetime performance traits in predicting total lifetime milk yield of holstein cows in egypt

الملخص EN

Data used in this study comprised 2730 lactation records of 850 Holstein cows sired by 316 sires.

The Holstein cows belong to a commercial farm.

The objective of this work was to determine to what extent total lifetime milk yield (TLM) is influenced by lifetime variables (total milk yield at first lactation (TMY1),total milk yield at last lactation (TMYL ), 305 milk yield at first lactation (305M1), 305 milk yield at last lactation (305ML), milk per day at productive life (Mday), number of complete lactations (NCL), lifetime days in milk (LDIM),productive life (Plife), age at disposal (CULL) and longevity index (LI, %) using the Partial Least Squares regression (PLS) procedure.

The Q² cumulated index (0.971) measures the global goodness of fit and the predictive quality of the TLM model.

The variables important for the projection of the total lifetime milk yield (TLM) were thoses measured-in-time (day or month or lactation) e.g.

LDIM, Plif , Cull, LI and NCL, but the variables measured-in-kg , e.g.

MDAY, TMYL, 305M1, 305ML and TMY1, had low influence on the TLM model .

The R² between the input variables (TLM and lifetime variables) and the PLS components was 0.97.

Therefore the model is well fitted.