Predict The Stock Price Using Principle Component Analysis And Neural Network

With the establishment of the stock market,it provides great support for economic construction and social development.More and more people are also involved in the stock investment.Many people want to predict the stock price correctly,which can get high returns,low risk.However,how to establish an accurate and effective stock forecasting model is of great significance for investors.On the basis of analyzing the theoretical knowledge of the stock market,the relevant problems of stock market prediction and the traditional stock forecasting methods,this paper proposes a combinatorial prediction method of a BP neural networks and principal component analysis.Firstly,the principal component analysis is performed by using SAS,which realize the dimension reduction of original data.Then,the selected principal components are used instead of the original input variables as the input variables of the BP neural network.So as to simplify the topology of the network,improve the convergence and stability of the neural network,and improve the generalization ability of the network.This experiment chooses the data of KAIDI stock as the original data.The results show that the BP network based on the improved principal component analysis through learning and training,which can forecast the stock data fitting and get good results.