Research on Stock Price Prediction Based on Orthogonal Gaussian Basis Function Expansion and Pearson Correlation Coefficient Weighted LSTM Neural Network

For stock price prediction in quantitative finance, deep learning techniques such as LSTM neural network do not need the stationarity assumption of traditional time series models (such ARIMA and GARCH) can forecast medium long-term series, so they have attracted much attention. This paper proposes an improved based on orthogonal Gaussian basis function expansion Pearson correlation coefficient weighting. The proposed method uses functional features intra-day prices to fit residual predicted by network. Considering that underlying model structure between each component eigenvector is unknown, we use Bagging capture trade off variance bias model. In addition, since dimension predictive variable a parameter be estimated, averaging weighting for tuning. results actual data analysis show significantly improve accuracy original has certain robustness. Finally, further applied consumer index (CPI) prediction, daily average temperature real-time monitoring environmental trace elements.