CEEMDAN分解下基于NAR神经网络的比特币价格预测

Bitcoin Price Prediction Based on NAR Neural Network under CEEMDAN Decomposition

为了帮助投资者更好地对比特币未来走势进行判断,利用CEEMDAN分解方法对比特币价格进行分解,并用NAR神经网络模型和ARIMA模型进行预测,同时与直接使用两个模型的预测效果进行比较。结果发现,在中长期上通过CEENDAN进行分解后预测精度更高,但运用在短期上会使精度降低,同时与ARI-MA模型相比,NAR神经网络模型的预测精度更高。

Bitcoin,the world's first digital currency which takes up a major share of the global digital currency market value,attracts a vast number o£investors.To help investors better compare the future of the special currency,the CEEMDAN decomposition method is used to decompose the Bitcoin price,the NAR neural network model and the ARIMA model are used to predict the price.And then the prediction effects of the two models are compared.The results show that the prediction accuracy is higher after decomposition by CEEMDAN in the medium and long term,while lower in the short term.Besides,compared with ARIMA model,the prediction accuracy of NAR neural network model is higher.