Comparative Analysis of Artificial Neural Network Algorithms for Prediction of FL305DMY in Murrah Buffalo
DOI:
https://doi.org/10.5455/ijlr.20200704062936Keywords:
Artificial Neural Networks, Bayesian Regularization, Murrah Buffalo, Test Day Milk YieldsAbstract
In the present study, first lactation records of 301 Murrah buffaloes were collected at Indian Veterinary Research Institute, Izatnagar. The first lactation 305-day milk yield (FL305DMY) of Murrah buffaloes was predicted using three different Artificial Neural Network (ANN) algorithms and their performance was compared. Performance of each algorithm was evaluated and compared on the basis of coefficient of determination and Root Mean Square Error (RMSE). In the study two different input sets were used to predict the milk yield. First input set consisted of four test day milk yields record along with Age at first calving (AFC) and Peak yield (PY) and second input set consisted of four monthly milk yields record, AFC and PY. ANN was trained using three algorithms that were Levenberg-Marquardt (LM), Bayesian Regularization (BR) and Scaled Conjugate Gradient (SCG). Each algorithm was compared using four different training-test data sets (66.66:33.33, 75:25, 80:20 and 90:10). BR achieved highest accuracy of 81.98% with lowest RMSE value of 16.13 % for input set-1 and 78.33% accuracy with RMSE value of 16.89 % for input set-2. Higher accuracy and lower RMSE value for BR algorithm clearly indicates it outperform SCG and LM algorithm. Hence, for prediction of FL305DMY in Murrah buffaloes by ANN should be done using BR algorithm as it shows better accuracy than LM and SCG algorithm.
Downloads
Published
How to Cite
Issue
Section
License
This work is licensed under a Creative Commons Attribution 4.0 International License.