Prediction Model of Chick Hatching Weight with Restricted Ridge Regression

Abstract

In this study, it is aimed to obtain prediction model for chick hatching weight by using shell ground color, shape characteristics, shell characteristics, white + yellow weight and weight loss property values during hatching of quail hatching eggs. Multicollinearity is one of the most common problems in prediction models. It is defined as linear dependence between explanatory variables in linear regression analysis. In this case, the ordinary least squares (OLS) estimator is not a predictor with minimum variance between unbiased estimators. Accordingly, the parameter estimates are away from the real value. One proposed approximation approach to eliminate this problem is restricted ridge regression.
 According to the findings obtained from 523 chick hatching eggs, elongation, egg shell weight, shell weight + embryo waste (g), absolute weight loss after transfer (g), relative weight loss after transfer (%) and absolute weight loss in 0-14 days (g) were determined as factors affecting chick hatching weight. In the linear modeling approach, which is used to determine these factors, multicollinearity problem is revealed. Therefore, in order to obtain a stable set of parameters, restricted ridge regression analysis was applied and estimations with lower standard errors than standard errors of the parameters obtained from the (least squares) LSQ approach were obtained.