THE USE OF THE PARTIAL LEAST SQUARES REGRESSION PROCEDURE TO PREDICT THE NET PROFIT IN PRODUCTION OF EGG-TYPE PULLETS

Abstract

SUMMARY Data on the financial variables of 43 cycles of Lohman Brown commercial eggtype pullets production were collected for 15 years (1990-2005) from Al-Nahda farm at Nubaria region in the north western part of Egypt. The flock size per cycle ranged between 25700 to 79500 birds, with a mean of 39600 birds, during the data collection period. The data were analyzed by the Partial Least Squares Regression (PLS) procedure using the XLSTAT 1.01, 2009 software. The objective of the study was to find out the possibility of projecting the net profit per-cycle (NP) as a dependent (response) variable from information on five independent (explanatory) cost variables. The cost variables included feed cost (FC), purchased chicks price (CP), veterinary cost (VC), cost of depreciation of the buildings and equipment (DC), and other cost (OC). These five variables were assembled in a principal component which was used in building the projection model (equation). Mean NP was L.E 62973, and the highest cost variable was FC (L.E 156182) representing about 46 % of the total cost. The correlation matrix between all variables showed significant correlation coefficients (P<0.05) between the independent variables and between each of them and the dependent variable (NP). The model quality index showed high explanatory power of the generated component for NP. An equation for the projection model of NP was built, and the goodness of fit statistics for NP showed high accuracy (R 2 = 0.90). The model can be