Comparative study of ANN and Regression methods for Prediction of Breeding Efficiency in Hariana cattle

Abstract

The present investigation was undertaken to predict the breeding efficiency by Tomar method (BET) using early life reproduction traits. Under this study, multiple linear regression (MLR) and artificial neural network (ANN) approach were used. Effectiveness of both methods was also compared for prediction of BE in Hariana cattle. Data of breeding information were collected from the history sheet registers and herd inventory registers of Hariana cattle maintained at DUVASU farm, Mathura. Data on productive animals born between 1962 and 2024 were gathered. Total 743 animal records were collected from Farm. The overall least-squares mean for breeding efficiency was estimated as 65.1 ±0.39%. Significant differences were observed between the periods of birth, AFC classes and parity of cows for BE. The heritability estimates for BET was 0.586. Breeding efficiency was predicted by multiple linear regression analysis. In MLR study, it was observed that model 5 having two independent variable (AFC and FSP) fulfilled most criteria such as higher R², and lower MSE, RMSE, MAE, MAPE, and U values. In the present investigation, the accuracy of prediction obtained from ANN was more as compared to MLR for prediction of BET using early life reproduction traits in Hariana cattle. The MLR explained 69.6% of accuracy of prediction of BET in Hariana cattle with MSE of 42.20. With the ANN algorithm, the MSE was found to be 33.3, which is more precise as compared to MLR. For the prediction of BET, MLR method is simple and easy compared to ANN, but accuracy of ANN was more.