This paper applied the wavelet method and support vector regression (SVR) to the oil price forecasting. We decomposed the non-steady time series of oil price into the long-term trend and the random disturbance in short time by using wavelet analysis method. Then the forecast of the crude oil price long-term trend was modeled by SVR approach. This paper considered multi-factors that could impact on the oil market, such as the supply and demand fundamentals, oil inventories, economy effects and market speculator activities etc. The empirical research showed that SVM model could forecast oil price long-run trend had high predictive performance, compared to the econometric regression method.%本文将小波分析与支持向量回归结合应用于国际原油价格预测,通过小波多尺度分析方法将油价时间序列分解为长期趋势和随机扰动项,然后采用支持向量回归对分解后的油价长期趋势进行预测.油价长期趋势的预测采用多因素预测方法,主要考虑市场供需基本面、库存、经济、投机等因素对石油价格走势的影响,建立多输入单输出的支持向量回归模型.实证研究表明,支持向量回归模型具有较高的预测性能,对原油价格长期趋势预测中,该方法比回归方法的预测精度高.
展开▼