A PCA-LSTM Model for Stock Index Prediction
Abstract
This paper proposed a LSTM network model to predict stock index closing price. During the research process, we noticed the multicollinearity of the variables in the volume-price information and solved it by using PCA principal component analysis. A stock index closing price forecasting model based on historical stock index price and volume-price information was constructed. The empirical study of the CSI300 index data in the model shows that the generalization ability of LSTM network after PCA processing is better than that of normal LSTM network and normal BP network. The prediction of training set and test set is more reliable.
Keywords
Financial market, LSTM model, Principle component analysis, Stock prediction model
DOI
10.12783/dtetr/ecar2018/26419
10.12783/dtetr/ecar2018/26419
Refbacks
- There are currently no refbacks.