Research On Investment Value Analysis And Predictive Modeling Of Stock Information

The stock market is not only a market with interests,but also a market with high risks.Making accurate forecasts of stock trends is of great significance to China's economic development and financial construction.Stock prices are affected by many factors,including politics and culture in macro factors,including industrial and regional factors,corporate factors and market factors,thus causing the complexity and instability of stock price forecasts.The stock market has a wealth of data resources,hiding a lot of information for mining.Data mining is to extract some potential rules that seem to violate common sense and actually reason in a variety of massive data,and use it to predict what humans can't predict and will happen in the future.and make the corresponding decision.And data mining technology has been widely used in finance,retail,manufacturing,healthcare and science industries.The analysis and prediction of stock issues is an important application area.This paper uses some indicators and data commonly used in the stock market.Using the Support Vector Machine,Random Forest and BP Neural Network in the data mining algorithms to analyze and predict the stock problems,the main contents are as follows:(1)Because the financial data of listed companies have the characteristics of high dimensionality and redundancy,combined with the data of some ST and non-ST companies,this paper proposes three machine learning algorithms to classify listed companies on the basis of Random Forest method for feature selection,in order to better judge the financial situation of listed companies and achieve the effect of early warning.(2)Among the mixed stocks,how to select high-quality stocks for scientific investment and obtain satisfactory returns has always been an important link in the process of investment value analysis.This paper puts forward the method of combining Analytic Hierarchy Process with Fuzzy Comprehensive Evaluation Method to evaluate the stock value objectively.(3)Aiming at the nonlinear and non-stationary problems of stock price data,a stock price forecasting method based on Principal Component Analysis and Generalized Regression Neural Network is proposed.Using Principal Component Analysis to reduce the dimensions of stock price,and forecasting stock price based on General Regression Neural Network model.(4)Due to the classical linear mapping dimension reduction methods such as Principal Component Analysis and Linear Discriminant Analysis,the nonlinear problem of stocks cannot get good results.A stock price forecasting method based on Local Linear Embedding algorithm and BP Neural Network is proposed.In this paper,we use a variety of models and methods in data mining to analyze the stock price forecast,stock classification and stock value investment.The simulation results show that the proposed method can be applied to China's stock market relatively effectively,which provides a corresponding reference for investors to more accurately predict stock prices and better grasp the development of the stock market.