Production and consumption of oil are critical
factors for the economics of a number of countries, so the ability
to predict the behavior of its price is very important. There are
many works that explore various methods of machine learning:
SVM, ANN, kNN, as well as statistical methods ARIMA,
GARCH and other time series analysis tools. The literature
review shows the advantages of SVM and ARIMA methods
in forecasting financial markets. In this paper, a technique for
constructing predictive models based on these two methods is
described. Conclusion is made that SVM model outperforms
ARIMA, that make it a good candidate for the energy prices
return prediction.
Index Terms—financial time series, oil prices, support vector
machine (SVM), ARIMA, financial prices forecasting