A COMPARISON OF ARTIFICIAL NEURAL NETWORKS AND TRADITIONAL TIME SERIES MODELS: A PREDICTION OF THE OIL PRICE

Abstract

The oil price has strong influence on the global economy, affecting several economic and energy sectors. Therefore, the objective of this work was to estimate oil prices in daily and weekly periods. For this, Artificial Neural Networks (ANNs) and traditional prediction models applied Computational Intelligence. The ANNs is an appropriate nonlinear technique for predicting complex markets. The results were compared with two models, simple moving average (MMS) and exponential smoothing (SE). The approach by Artificial Neural Networks presented an average absolute error, smaller than the other techniques used. In addition, it was observed that this method allows using different variables to make future forecasts, for example, to carry out a certain forecast economic, environmental and demographic indicators can be considered.