Open AccessImplementasi Jaringan Syaraf Tiruan Perambatan Balik untuk Memprediksi Harga Logam Mulia Emas Menggunakan Algoritma Lavenberg Marquardt
Author(s) -
Reza Najib Hidayat,
R. Rizal Isnanto,
Oky Dwi Nurhayati
Publication year - 2013
Publication title -
jurnal teknologi dan sistem komputer
Language(s) - English
Resource type - Journals
eISSN - 2620-4002
pISSN - 2338-0403
DOI - 10.14710/jtsiskom.1.2.2013.49-55
Subject(s) - artificial neural network , investment (military) , layer (electronics) , investment value , value (mathematics) , computer science , mathematics , artificial intelligence , statistics , business , finance , materials science , composite material , politics , political science , law , cash
Gold is one of commodities investment which its value continue to increase by year. The rising price of gold will encourage investors to choose to invest in gold rather than the stock market. With the risks that are relatively low, gold can give better resultsin accordance with its increasing price. In addition, gold can also be a safe value protector in the future.The Objectives of the research are to predict the price of gold using artificial neural networks backpropagations methods and to analyze best network used in prediction. In the process of training data, it is used some training parameters to decide the best gold prediction architecture. Comparative parameters that is used are the variation of the number of hidden layers, number of neurons in each hidden layer, learning rate, minimum gradients and fault tolerance. The results showed that the best architecture has an accuracy rate of 99,7604% of data training and test data at 98,849% with architecture combinations are have two hidden layer neurons combined 10-30, the error rate 0.00001 and 0.00001 of learning rate.