Model of Artificial Neural Networks in Predictions of Corn Productivity in an Effort to Overcome Imports in Indonesia
Anjar Wanto, Dedy Hartama, Gita Widi Bhawika, Zulifah Chikmawati, Deswidya Sukrisna Hutauruk, Pinondang Hotria Siregar, Ricard Fredrik Marpaung, Salim Efendi, Imeldawaty Gultom, Agus Perdana Windarto · Journal of Physics Conference Series · 2019
Abstract Indonesian people still consume a lot of corn as a staple food. Corn productivity based on data from the Indonesian Central Bureau of Statistics from 2005 to 2015 is not stable. Therefore this study was conducted to determine the right prediction model to see the level of corn productivity in Indonesia for the coming year, with the hope that the government has a reference to continue to work to improve corn productivity to remain stable in order to meet the needs and minimize corn imports. This study uses data on corn productivity in 2005-2015 sourced from the Indonesian Central Bureau of Statistics. The algorithm used to determine the prediction model is the Backpropagation artificial neural network. This algorithm is able to predict times series data. Based on this algorithm, the training and testing process is carried out using 5 network architecture models, namely 5-25-1, 5-43-1, 5-76-1, 5-78-1 and 7-128-1. The best architecture obtained from the 5 models is 5-25-1 with 88% accuracy percentage and MSE value 0, 00992433.