黄金价格的小波分解与单支重构求和预测

Summation prediction of gold price based on wavelet decomposition and single reconstruction

针对差分法会造成时间序列信息损失的缺陷,首先对2000年1月至2013年4月黄金价格(月度)序列进行了小波分解和单支重构,其次对重构的近似分量和细节分量分别建立非参数自回归模型和AR模型,最后对每个分支进行了14步外推预测,将这些预测值求和得到了原始黄金价格序列自2013年5月至2014年6月的预测值。预测结果表明:2013年5—12月,黄金价格稳中有降,但降幅较小;进入2014年后,黄金价格则稳中上升,但上升幅度较小。

In view of the defects of difference causing information loss of time series, firstly, wavelet decomposition and single reconstruction are carried out for international gold monthly price series from 2000 January to 2013 April in the paper. Secondly, nonparametrie autoregressive model and AR model are respectively established for the approximation and detail components; Finally, through summation of the prediction values, the prediction values of original gold price series are obtained from May,2013 to June,2014 based on the 14 steps extrapolation prediction for each branch. The results show that during the period from May to December,2013, gold prices are overall steady with a slight fall; entering 2014,gold prices keep undisturbed except by healthy growth,but in a minor way.