All Rights Reserved
AccessEcon LLC 2006, 2008.
Powered by MinhViet JSC
ralph lauren polo

Manel Hamdi, Chaker Aloui and Santosh kumar Nanda
''Comparing Functional Link Artificial Neural Network And Multilayer Feedforward Neural Network Model To Forecast Crude Oil Prices''
( 2016, Vol. 36 No.4 )
In this paper a trigonometric functional link artificial neural network (FLANN) model using backpropagation rule is applied to predict the next day's spot price of US crude oil. The daily observations of these variables: US dollar index, S&P 500 stock price index, gold spot price, heating oil spot price and US crude oil spot price are employed as inputs of the proposed model. By comparing with multilayer backpropagation feedforward neural network (FNN), more accurate predictions were shown by applying the FLANN model. In fact, several performance criteria are used to assess the forecasting power of the proposed model such as the Root Mean Squared Error (RMSE), the Mean Absolute Error (MAE) and the hit rate. For checking the forecasting robustness of the proposed model, in addition to the other input variables, the US crude oil and biofuels production are also used to predict the next month's spot price of crude oil. Comparatively, similar conclusion was deduced and the FLANN model performs better than the standard FNN. These findings can be explained by the simplicity of FLANN structure since it consists of a single layer with only one neuron at the output thus a lower computational load on the network.
Keywords: Crude oil price, Forecasting, Functional link artificial neural network (FLANN), Multilayer feedforward neural network (FNN).
JEL: C5 - Econometric Modeling: General
Q4 - Energy: General
Manuscript Received : Feb 21 2016 Manuscript Accepted : Dec 10 2016

  This abstract has been downloaded 195 times                The Full PDF of this paper has been downloaded 99548 times