Predictive ability of egg production models

Abstract

The monthly egg production data of a strain of Rhode Island chickens were used to compare three mathematical models (the Parabolic exponential, Wood's Gamma and modified Gamma by McNally) on their ability to predict 52 week total egg production from part-production at 16, 20, and 24 weeks, on a hen-housed basis. The results suggest that the three models fitted 52 week laying records quite well, judging from their respective R values, which were higher for McNally (0.95) and Parabolic exponential (0.93) than for Wood (0.75). However, their ability to predict 52 week egg production from part-records of 16, 20 24 wks varied. The prediction of total production based on fit to 24 week of data was more accurate for the McNally. The latter consistently predicted less than the actual 52 week (-1.13% deviation), but the other 2 models tended to over estimate production (i.e. 12.26 and 13% deviations for Parabolic exponential and Wood models, respectively). It was concluded that based on the goodness of fit to 52 week production record and accuracy to predict full record egg production from part record, the McNally model gave the best results, and could therefore be said to have theoretical advantages over the other models. It may thus be found useful in decision making concerning replacement of layer flocks.