变系数模型在国际油价预测中的应用  

Application of varying-coefficient regression model in international oil price forecasting

作　　者：王艳[1] 汪金菊[1] 

机构地区：[1]合肥工业大学数学学院,安徽合肥230009

出　　处：《阜阳师范学院学报（自然科学版）》2014年第1期11-15,共5页Journal of Fuyang Normal University(Natural Science)

基　　金：安徽省自然科学基金项目(1208085MF91;11040606M03);教育部人文社会科学研究项目(10YJA910005)资助

摘　　要：对影响石油价格的间接因素变量采用取对数的方式,建立无季节性与存在季节性虚拟变量的变系数模型。同时利用变系数模型对2001年第一季度到2012年第一季度的西德克萨斯中质原油(WTI)现货价格序列进行预测比较分析。实证结果表明,建立的无季节性虚拟变量的变系数模型能够更有效地预测石油价格序列,同时它的预测效果比直接运用历史数据的效果更好。最后通过拟合优度系数的计算可知对石油价格有影响因素的选择是正确的。Utilizing logarithms for indirect factor variables that influence the oil price, the study built a varying-coefficient regression model of no seasonal and existing seasonal dummy variables. And then the models were used to forecast, eompare and ana- lyze the spot price sequence of West Texas intermediate （WTI） crude oil from the first quarter of 2001 to the first quarter of 2012. The result showed that the model of no seasonal and dummy vm＇iable can more efficiently forecast oil price sequence and its forecast effect is better than that of the direct factor variables. Finally, the counting of the goodness-of-fit coefficient proved that we have chosen the right influential factors to forecast the oil price.

关 键 词：变系数模型 加权最小二乘法 石油价格 预测 

分 类 号：O212.1[理学—概率论与数理统计]