Artificial Neural Network For Estimation Nutrient Utilization Based on Chemical Composition on Ruminant Animal Feed

The production of ruminant livestock is used to meet the needs of human food. The productivity of ruminants is determined by the feed given. Therefore, a proper feeding strategy is needed. This research aims to estimate the nutrient utilization of ruminant forage plants based on chemical content using Artificial Neural Network (ANN). Chemical content used as the estimator is the main analysis category; meanwhile, ruminant nutritive values are predicted from our model. A total of 136 species of forage plants were obtained from http://www.feedipedia.org. Data were divided into training and testing data using K-fold cross-validation with k=3. The process of estimation was performed using the ANN structure with several variation numbers of neurons in the hidden layer from 4, 6, and 8. As a result, the ANN structure built could estimate the nutrient properly, indicated by small values of RMSE (3.33) revealed from ANN structure with four neurons. The determination coefficient (R²) from this ANN model more than 0.50 for all output, except in the production of Metabolizable energy, which was 0.49.