Analysis of the Accuracy Batch Training Method in Viewing Indonesian Fisheries Cultivation Company Development
Anjar Wanto, Parulian Parulian, Hasudungan Siahaan, Agus Perdana Windarto, Teuku Afriliansyah, Widodo Saputra, Iin Melya Parlina, Muhammad Ridwan Lubis, Irfan Sudahri Damanik · Journal of Physics Conference Series · 2019
Analysis of research is very important to do, so that research becomes more precise and directed. As well as analyzing the development of aquaculture companies in Indonesia, studies and the use of appropriate methods are needed to get optimal results. This research is expected to be widely useful, both for the Government of Indonesia and the private sector as one of the study materials in business development in the fisheries sector. The data used in this study is data on the number of aquaculture companies according to the type of cultivation obtained from the Indonesian Statistical Center from 2000 to 2016. This study uses the Batch training method with weight and bias learning rules with 5 architectural models, namely: 8-5-1, 8-10-1, 8-5-10-1, 8-10-20-1 and 8-20-40-1. Of these 5 architectural models, the best is 8-5-1 with an accuracy rate of 75%, MSE 0.0445464533, with an error rate of 0.001 - 0.07.