APPLICATION ON STOCK PRICE PREDICTION OF ELMAN NEURAL NETWORKS BASED ON PRINCIPAL COMPONENT ANALYSIS METHOD

Study on the prediction of stock price has great theoretical significance and application value. Traditional stock forecasting methods cannot fit and analysis highly nonlinear, multi-factors of stock market well, there are problems such as the prediction accuracy is not high, the slow training speed etc. In order to improve the accuracy of stock price forecasting, this paper proposes a prediction method of Elman neural network model based on principal component analysis method. In order to better compare results, establish structure same BP network and Elman network, forecast for stock data; then using principal component analysis filter factors of significant effect on stock prices, Elman neural network model based on principal component analysis method, and compared with single Elman network and BP networks prediction results. Result shows BP network convergence is relatively slow, train for a long time, and could converge to a local minimum; Elman network training time is short, the error bars for smoother and more stable performance; Elman neural network model based on principal component analysis with higher accuracy, faster network speeds.