基于振荡序列的黄金价格动态建模及预测

Dynamic modeling and forcasting based on the gold price oscillation sequence

对黄金价格预测时常采用大样本数据建模方法,该方法首先对数据进行统计分析,找到数据内在统计特征,再寻求符合该特征的模型;但该方法在数据统计分析时受到统计手段和方法的约束,数据本身的内在特征往往不能充分反映,因此模型的选择受到限制。针对黄金价格数据不具有趋势性变化的特征,利用灰色GM(1,1)系统建模理论、数据驱动建模原理及振荡序列的数据处理方法,建立了基于振荡序列的黄金价格动态预测模型。该模型不仅对黄金价格数据自身存在的振荡性进行了处理,避免了数据大波动,同时利用动态数据驱动方法不断更新数据,提高了预测精度。

Forecast of gold prices often adopts large sample data modeling method. The method starts with statistical analysis of data to find the internal statistical characteristics,and then seeks to conform to the characteristics of the model. But this method is constrained by the statistical means and method during statistic analysis of data so that the internal characteristics of the data themselves can not be fully reflected. So the selection of models is limited. In view that gold price data have no trend change characteristics,the paper uses the theory of grey system GM（ 1,1） modeling,metabolic modeling principle and oscillation sequence data processing method,and sets up the gold price dynamic prediction model based on sequence of oscillation. The model not only deals with the oscillation behavior of the data to avoid great fluctuation,but at the same time improves the prediction accuracy by using dynamic data driving to update data.