Author(s)

Jing Ren

Corresponding Author

Jing Ren

ABSTRACT

This paper takes the wavelet analysis into the prediction of BP and LSTM neural networks, and selects the closing price of the Shanghai Composite Index (000001) from July 1, 2016 to June 30, 2021. In order to obtain the yield, this paper takes the first-order difference of the natural logarithm. Through wavelet analysis, the original yield sequence is decomposed and reconstructed. Then, the BP and LSTM neural network models are constructed with the reconstructed sequence, and the prediction effect of the two models is compared and analyzed. The result shows the fitting effect of LSTM is better than that of BP neural network. Further, the BP and LSTM neural network models of the original yield sequence are compared with the wavelet reconstructed model, and it is found that the wavelet decomposition and reconstruction can improve the accuracy of prediction.

KEYWORDS

Wavelet Analysis, BP Neural Network, LSTM Neural Network