
基于混合多分形小波的国际油价多期预测研究
Multi-Period Prediction of Oil Price with a Hybrid Multifractal Wavelet Model
本文综合Haar小波和乘性级联两种树型结构的优势构建一种混合多分形小波模型(H-MWM)来对国际油价进行多期预测.首先,对日度油价做Haar小波三层分解,提取粗粒度层(尺度系数)数据,对尺度系数做单步预测;其次,将日度油价做乘性级联三层分解,提取各层的细粒度(乘子)数据,对各层的乘子做预测;然后,构建尺度系数与乘子间数量关系,用预测的尺度系数和乘子,得到各层小波系数预测值;最后,将尺度系数和小波系数预测值,通过Haar小波重构为原序列粒度,得到日度油价多期预测值.实证研究表明:构建的H-MWM方法在日度油价的多期预测中,在保证预测准确度提高的同时,大大降低了计算时间复杂度.
This paper combines the advantages of Haar wavelet and multiplicative cascaded tree structure to build a hybrid multifractal wavelet model (H-MWM) for multi-period prediction of international oil price. First, the daily oil price is decomposed into three layers using Haar wavelet, and the coarse-grained layer (scale coefficient) data is extracted. Then the scale coefficient is predicted in one step. Second, the daily oil price is decomposed into three layers using the multiplication cascade method, and the fine-grained (multiplier) data of each layer is extracted. Then the multipliers of each layer are predicted. Third, the quantitative relationship between scale coefficient and multipliers is constructed. Using the predicted scale coefficient and multipliers, the prediction values of wavelet coefficients in each layer are obtained. Finally, the prediction values of scale coefficient and wavelet coefficient are reconstructed into the original sequence granularity through Haar wavelet, and accordingly the multi-period prediction values of daily oil price are obtained. Empirical results show that the proposed H-MWM method can not only improve the prediction accuracy, but also reduce the computational time complexity.
多重分形 / Haar小波 / 树型结构 / 多期预测 / 油价预测 {{custom_keyword}} /
multifractal / Haar wavelet / tree structure / multi-period prediction / oil price forecasting {{custom_keyword}} /
表1 预测序列评价标准 |
计算步骤 | 方向精度 | 水平精度 |
1. 每次得到8日预测序列 | ||
2. 8日的精度取平均 | ||
3. 将测试集的平均误差组合成一条平均误差序列 |
表2 |
Test models | Benchmarks_Dstat | Benchmarks_MAPE | |||||
Haar | Cascade | Roll | Haar | Cascade | Roll | ||
H-MWM | 0.00000 | 0.00000 | 0.00000 | 0.00000 | 0.00000 | 0.00005 | |
Haar | 0.17724 | 0.26987 | 0.00000 | 0.95567 | |||
Cascade | 0.61911 | 1.00000 |
注: |
梁强, 范英, 魏一鸣, 基于PMRS的期货加权油价多步预测方法[J]. 管理科学学报, 2008, 11 (6): 84- 90.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
尚永庆, 王震, 陈冬月, Hull-White模型和二叉树模型在预测油价及油价波动风险上的应用[J]. 系统工程理论与实践, 2012, 32 (9): 1996- 2002.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
唐振鹏, 张婷婷, 吴俊传, 杜晓旭, 陈凯杰, 基于混合模型的原油价格多步预测研究[J]. 计量经济学报, 2021, 1 (2): 346- 361.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
王升辉, 裘正定, MPEG-4视频流量多重分形建模[J]. 通信学报, 2006, (10): 44- 51.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
邬源杨, 董玮文, 杨宇航, 小波在长相关流量工程中的应用[J]. 通信技术, 2002, (9): 76- 77.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
Abdullah S N, Zeng X, (2010). Machine Learning Approach for Crude Oil Price Prediction with Artificial Neural Networks-quantitative (ANN-Q) Model[C]// Proceedings of IEEE the 2010 International Joint Conference on Neural Networks: 1-8.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
Abry P, Flandrin P, Taqqu M S, Veitch D, (2000). Wavelets for the Analysis, Estimation, and Synthesis of Scaling Data[M]// Self-Similar Network Traffic and Performance Evaluation. New York: Wiley-Interscience.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
{{custom_citation.content}}
{{custom_citation.annotation}}
|
{{custom_citation.content}}
{{custom_citation.annotation}}
|
{{custom_citation.content}}
{{custom_citation.annotation}}
|
{{custom_citation.content}}
{{custom_citation.annotation}}
|
{{custom_citation.content}}
{{custom_citation.annotation}}
|
{{custom_citation.content}}
{{custom_citation.annotation}}
|
{{custom_citation.content}}
{{custom_citation.annotation}}
|
{{custom_citation.content}}
{{custom_citation.annotation}}
|
{{custom_citation.content}}
{{custom_citation.annotation}}
|
{{custom_citation.content}}
{{custom_citation.annotation}}
|
{{custom_citation.content}}
{{custom_citation.annotation}}
|
{{custom_citation.content}}
{{custom_citation.annotation}}
|
{{custom_citation.content}}
{{custom_citation.annotation}}
|
{{custom_citation.content}}
{{custom_citation.annotation}}
|
{{custom_citation.content}}
{{custom_citation.annotation}}
|
{{custom_citation.content}}
{{custom_citation.annotation}}
|
{{custom_citation.content}}
{{custom_citation.annotation}}
|
{{custom_citation.content}}
{{custom_citation.annotation}}
|
{{custom_citation.content}}
{{custom_citation.annotation}}
|
{{custom_citation.content}}
{{custom_citation.annotation}}
|
{{custom_citation.content}}
{{custom_citation.annotation}}
|
{{custom_citation.content}}
{{custom_citation.annotation}}
|
{{custom_citation.content}}
{{custom_citation.annotation}}
|
{{custom_ref.label}} |
{{custom_citation.content}}
{{custom_citation.annotation}}
|
/
〈 |
|
〉 |