Linear regression for birth weight of chicks in function of egg weight of naturalized chicken without defined ecotype
A great concern of country chicken breeders is the chick sizes, which can be influenced by the egg's weight (Paiva et al. 2010). Thus, the purpose of this study was to fit a generalized linear model (GLM) to predict the weight of chicks in function of morphometric measurements of eggs. The present study was carried out at the aviary of Embrapa Meio-Norte. 224 eggs were used, in which following measurements were performed: weight at incubation (WI), length (LE) and width (WE) of eggs, birth weight of chick (BW) and weight of shell (WS). The analysis was carried out by the R software (version 3.2.1). The Anderson-Darling normality test (p-value=0.05) was applied on the birth weight of chicks. So, the normal distribution was used to fit the GLM with the link functions (identity, logarithmic and inverse). In the three scenarios, the variables LE, WS and WI were significant, while WE was not significant. Exclusion of WE probably occurred due to a negative correlation between this variable and LE. Residual analysis was used to select the model, since the classical criteria (AIC, BIC and R2) are penalized by the amount of parameters (the same in the 3 models), they were not relevant. Residues of all models were normally distributed according to the Anderson-Darling test (p-value=0.05). On avarage, residues are null on order of 10-6. However, the model with logarithmic link function was selected, since its variability of was lower than to the other models. The predicted model for birth weight of chicks was denoted as BW = 2.816 - 0.005*LE + 0.007*WS + 0.016*WI. The length of egg negatively affects the weight of chick in 0.005 for each unity that was added. On the other hand, the weight of shell and the weight at incubation positively influenced birth weight of chicks in 0.007 and 0.016 for each unity of WS and WI that was added, respectively. An advantage of this technique is the possibility to predict the birth weight of chick based on the independent variables, for example, considering LE = 50, WS = 4 and WI = 45, BW will be exp {2.816 - 0.005*(50) + 0.007*(4) + 0.016*(45)} = 27.446.