Artificial Neural Network versus Multiple Regression Analysis for Prediction of Lifetime Milk Production in Sahiwal Cattle

Neural network and Bayesian-based prediction of breeding values in Beetal goat

The estimation of breeding values is prime concern for animal breeders in order to achieve desired genetic progress of farm animals. However, current methods for estimating BV involve simultaneous selection of animal model which are computationally intensive and time-consuming. The present attempt was made to predict breeding values of weaning trait under artificial neural networks (ANN), Bayesian technique (BT), and multiple regression (MR) methods. The data records comprising year of birth, sex, type of birth, dam's weight at kidding, birth weight, weaning weight, and estimated breeding values (BV) for weaning weight (under animal model) pertaining to 849 kids born to 37 sires and 237 dams between 2004 and 2019 were used in this study. All three methods, viz., ANN under multilayer perceptron (2 hidden layers), BT under Markov chain Monte Carlo (MCMC) approach, and MR under full model, were used for 75% training dataset initially and prediction model developed was applied on 25% test dataset. The initial analysis showed positive and significant (P < 0.01) relationship of BV with other variables which hinted that BV may be predicted with accuracy. Then, it was revealed from the results indicated that ANN, BT, and MR methods have similar accuracy (r = 0.86 to 0.87) for prediction of BV. However, ANN showed slightly higher but negligible model adequacy than BT and MR method. The prediction error under three methods was almost equal. The results indicated that these methods could be used as potential alternative for recurrent prediction of BV based on phenotypic data in order to optimize selection plans at young age in resourced population of Beetal goat.

Visible and Near-Infrared Multispectral Features in Conjunction with Artificial Neural Network and Partial Least Squares for Predicting Biochemical and Micro-Structural Features of Beef Muscles

The objective of this study was to determine the potential of multispectral imaging (MSI) data recorded in the visible and near infrared electromagnetic regions to predict the structural features of intramuscular connective tissue, the proportion of intramuscular fat (IMF), and some characteristic parameters of muscle fibers involved in beef sensory quality. In order to do this, samples from three muscles (Longissimus thoracis, Semimembranosus and Biceps femoris) of animals belonging to three breeds (Aberdeen Angus, Limousine, and Blonde d’Aquitaine) were used (120 samples). After the acquisition of images by MSI and segmentation of their morphological parameters, a back propagation artificial neural network (ANN) model coupled with partial least squares was applied to predict the muscular parameters cited above. The results presented a high accuracy and are promising (R² test > 0.90) for practical applications. For example, considering the prediction of IMF, the regression model giving the best ANN model exhibited R²P = 0.99 and RMSEP = 0.103 g × 100 g⁻¹ DM.

Comparative study of ANN and conventional methods in forecasting first lactation milk yield in Murrah buffalo

Present investigation was undertaken to predict first lactation 305-day milk yield (FL305DMY) using monthly test day milk records. 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 FL305DMY in Murrah buffalo. The data on 3336 monthly test day milk yields records of first lactation pertaining to 556 Murrah buffaloes maintained at National Dairy Research Institute, Karnal; Central Institute for research on buffalo; Guru Angad Dev Veterinary and Animal Sciences University (GADVASU), Ludhiana and Choudhary Charan Singh Haryana Agricultural University (CCSHAU), Hisar were used in this study. In MLR study, it was observed that model 14 having four independent variable, i.e. FSP, TD2, TD4 and TD6 fulfilled most criteria such as highest R2, lowest MSE, lowest RMSE, lowest CP, lowest MAE, lowest MAPE, and lowest U value. In the present investigation, the accuracy of prediction obtained from ANN was almost similar to MLR for prediction of FL305DMY using monthly test day milk records in Murrah buffalo. The best ANN algorithm achieved 76.8% accuracy of prediction for optimum model, whereas the MLR explained 76.9% of accuracy of prediction of FL305DMY in Murrah buffalo. MLR method is simple as compared to ANN, hence MLR method could be preferred.

Comparison of the decision tree, artificial neural network and multiple regression methods for prediction of carcass tissues composition of goat kids

The aim of this study was to predict carcass tissue composition of goat kids using the decision tree with CHAID algorithm (DT) and artificial neural network (ANN) method in comparison with classical step-wise regression (SWR) analyse. Data were obtained from 57 goat kids of Gokceada breed. Predictor variables were pre-slaughter weight, several carcass measurements and indices, weights of different carcass joints and dressing percentage. R² values ranging from 0.212 to 0.371 indicating low to moderate accuracy were obtained for predicting muscle proportion. DT and ANN yielded similar R² values for predicting bone proportion. DT was the best prediction method for estimating proportions of subcutaneous fat (R² = 0.828) and intermuscular fat (R² = 0.789). According to DT analyses, cold carcass weight was the most important factor influencing bone proportion, while kidney knob and channel fat weight was the predominant factor influencing subcutaneous, intermuscular and total fat proportions. Consequently, the use of DT method can be considered to predict carcass fat proportions.