Open AccessForecasting with Feed Forward Neural Network model and adaptive simulated annealing algorithm (Case: world crude oil prices that was published by OPEC)
Author(s) -
A. Hanafie,
Sugito Sugito,
S. Sudarno,
Aradea R. Hakim
Publication year - 2019
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1217/1/012079
Subject(s) - artificial neural network , simulated annealing , computer science , crude oil , feedforward neural network , autoregressive integrated moving average , algorithm , mathematical optimization , time series , artificial intelligence , machine learning , engineering , mathematics , petroleum engineering
Today, crude oil trading industry became an important industry in the world, it was caused by highly fuel oil consumption. The price of crude oil has a changing price trend, it makes the prediction of crude oil in the coming periods to be challenging. Various methods can use to forecast price of crude oil, it is using ARIMA Box-Jenkins model with OLS method to estimate the parameter, but this method has several assumptions that must be complete. Overtime, many methods were develop, one of them is artificial neural network can be combine with various parameter optimization methods such as Adaptive Simulated Annealing algorithm. Adaptive Simulated Annealing algorithm is an optimization method it was inspired by the process of crystallization, the advantages of this algorithm has a running time faster than similar algorithms. The combination of artificial neural networks and Adaptive Simulated Annealing algorithms can be used to model the historical data without requiring assumptions in the analysis. Based on the analysis on this research, the best model is obtained FFNN 2-5-1 with MAPE value of 1.0042%.