Use of Data Mining Algorithms Chaid and Cart in Predicting Egg Weight from Egg Quality Traits of Indigenous Free-Range Chickens in Zambia

This study aimed to evaluate the performance of data mining algorithms Chi-square Automatic Interaction Detector (CHAID) and Classification an‎d Regression Tree (CART) in predicting egg weight from different egg quality traits such as egg length (EL), egg width (EW), shell weight (SW), shell thickness (ST), albumen weight (AW), yolk height (YH), yolk width (YD) and yolk weight (YW). To attain this, 364 indigenous free-range chicken eggs were employed. The goodness of fit test was done to compare the predictive performance of these algorithms. Pearson correlation coefficient between the egg weight and the predicted egg weights was found to be 0.907 (P < 0.01) for the CHAID algorithm. Coefficient of determination (R2 ), Adjusted R2 (Adj-R2 ), Root Mean Square Error (RMSE), Relative approximation error (RAE), standard deviation ratio (SD ratio) for the CHAID algorithm were estimated to be 0.823, 0.823, 2.23, 0.04, and 0.23 correspondingly. Pearson correlation coefficient between the egg weight and the predicted egg weights was found to be 0.771 (P < 0.01) for the CART algorithm. The R2 , Adj-R2 , RMSE, RAE, SD ratio for the CART algorithm were estimated to be 0.593, 0.593, 2.32, 0.07, and 0.24 correspondingly. Given its stronger prediction accuracy (R2 ), lower values of RMSE, and RAE, the CHAID algorithm can be recommended for analysis of egg quality traits of free-range indigenous chickens.