基于SVR的石油期货价格短期预测  

Petroleum Futures Prices’ Short-term Forecasting Based on SVR

作　　者：慕晓茜[1] 何佳[1] 倪旭敏[1] 段雨墨[1] 陆昊[1] 张作泉[1] 

机构地区：[1]北京交通大学理学院,北京100044

出　　处：《科学技术与工程》2010年第18期4585-4589,共5页Science Technology and Engineering

基　　金：大学生创新性实验计划项目资助

摘　　要：将SVR原理引入到石油期货价格的时间序列中,并以美原油价格进行了实证分析。结果表明:该方法能充分反映石油期货价格序列走势,对短期价格的预测具有较高的精度;并且还发现,超级参数的选择服从一定规律,即其乘积在一定范围内效果较佳。之后,还将此理论推广到多维影响因素和其他金融时间序列的预测中。The support vector regression （SVR） principle is introduced into the petroleum futures forward price and has carried on the empirical analysis by the US crude price. The results indicate that this method can fully reflect the petroleum forward price sequence trend and has the high precision to short term price′s forecast. Moreover,super parameter′s choice obedience certain rule,namely the majority of its product falls within a certain range. This theory is extended to the multidimensional impact of this factor and other financial time series also.

关 键 词：支持向量机回归 石油期货价格 时间序列预测 拐点预测 

分 类 号：F830.91[经济管理—金融学]