论文部分内容阅读
本文基于分解-重构-集成的思想,构建了一个多尺度组合预测模型,选取小麦作为粮食的代表,预测其价格走势。首先,运用集合经验模态分解方法(EEMD)分解价格序列,然后,用灰色关联分析方法对分量序列进行重构,重构为高频、中频、低频和趋势项四个部分,并从不规则因素、季节因素、重大事件和世界经济水平等方面对这四个部分波动特点进行解释,针对不同特点的分量选择不同的方法进行预测,最后对各预测结果用支持向量机集成,并与其他预测模型的预测结果进行比较。实证结果表明,本文构建的多尺度组合模型的预测效果优于灰色预测GM(1,1)、BP神经网络、SVM方法、ARIMA模型等单模型方法和ARIMA-SVM组合模型以及基于EMD和EEMD分解的其他多尺度组合模型。
Based on the idea of decomposition-reconstruction-integration, this paper constructs a multi-scale combined forecasting model, selects wheat as the representative of grain, and forecasts its price trend. Firstly, the set of empirical mode decomposition (EEMD) is used to decompose the price series. Then, the gray relational analysis method is used to reconstruct the component sequence into four parts: high frequency, intermediate frequency, low frequency and trend item. Factors, seasonal factors, major events and world economic level, explain the fluctuation characteristics of these four parts, choose different methods to predict the different characteristics of the components, and finally integrate the prediction results with support vector machines and compare them with other forecasts The model predictions are compared. The empirical results show that the proposed multi-scale combined model is better than gray prediction GM (1,1), BP neural network, SVM method, ARIMA model and ARIMA-SVM combination model, and EMD and EEMD decomposition Other multi-scale combination model.