基于BP神经网络的黄金价格非线性组合预测模型

摘要

For the prediction of the gold price, it is difficult for single model to fully reflect the variation regulation of the gold price. In order to make full use of the advantages of every model, different forecasting methods are combined to produce better prediction effect. The nonlinear combination prediction model of gold price is established by combining several single models with BP neural network. The prediction accuracy of the nonlinear combination model is higher than the single model and the linear combination model without co-integration function.%对黄金价格进行预测时,单一模型往往难以全面反映黄金价格的变化规律.为了更有效地利用各模型的优点,将不同的预测模型进行组合可以产生更好的预测效果.利用BP神经网络对单一模型进行非线性组合,建立了黄金价格的非线性组合预测模型.实证研究结果表明,非线性组合模型的预测精度高于被组合的单一模型和不具有协整关系的线性组合模型.