The Research Of The Application Of Neural Network In Short-term Stock Price Forecasting

With China’s rapidly economy growth and the stock market continues to expand,the stock market has generated a lot of valuable data. The investors use these data as animportant basis for analysis of stock investment. At the same time, Stock price forecasthas also become the research object of the investors. The data forecasting techniqueswere the important method in prediction. BP neural network attracted much attentionfrom the investors and researchers in the field of data prediction. On the other hand,some of the shortcomings of the BP algorithm are also inaccurate.In this paper, based on analysis of the problems faced by the short-term forecast ofthe stock price, the paper explored the principal component analysis, the feasibility ofthe genetic algorithm and BP neural network short-term forecast on the stock price. BPneural network can learn the stock market data to identify the inherent law ofdevelopment and changes in stock market. The main research work is as follows:This paper uses the principal component analysis method to solve the BP neuralnetwork input vector dimension reduction. At the same time, in order to improve theprediction accuracy, research combined the fitting and principal component analysisproposed the method named principal component analysis and written algorithm byMatlab7.0.This paper uses genetic algorithm to improve BP algorithm to overcome the localminimum defects and genetic neural network model to short-term forecast the stockprice. Based on the previous algorithm, this paper establishes a comprehensivepredictive model and successful implements in Matlab7.0.At last, in order to test the effectiveness of the algorithm, this paper uses thepreviously proposed algorithm to do the experiment. Then, error analysis is carried outon the experimental basis. The results show that the algorithm proposed by this paperis effectively.