Design And Application Of Stock Analysis And Forecast Based On Data Mining

As a weatherglass of macroeconomic and an important indicator of microeconomic, Stock market has been linked people’s economic life. So large investment risks in stock market that those investors when they make an investment decisions need to analysis a large number of stock market history trading data and financial data to select the investment direction. The human brain is limited capacity to process the data. However these huge amounts of data can be deal by Data Mining who make the data Analysis, found that the data model and features, a reasonable investment analysis and forecasting. Therefore, the data mining analysis and forecast in the application of the stock has important practical significance is beyond doubt.The paper expatiates how to analyze and forecast stock by the technical analysis aspect and fundamental analysis aspect. With the use of data mining techniques and integrated use of mathematics, finance and other related knowledge, the solution of specific problems in analysis and forecasting of stock was put forward, and make the corresponding Analysis research program.The main contents including:The stock trading data was compared and analyzed by using the neural network and Support vector machine. First, the neural network and support vector machine theory are introduced, and the advantages and disadvantages of both techniques were compared and analyzed. Then, The Shanghai Stock Index was predicted by using support vector machines and neural networks. Finally, the conclusion is that support vector machines is than the neural network of high value in the prediction accuracy at the short-term, but the neural networks in terms of the support vector machine to predict the long-term trends; support vector machine prediction results when the trend reverses Deviation is than the large neural network.Cross-industry standard process of data mining (CRISP-DM) was introduced. As financial data analysis and forecast system, generated through the use of C5.0 decision tree algorithm to select a representative financial indicators, analysis of high-growth companies and general financial indicators in those differences. The clustering technique was applied to the analysis of financial data on stocks, stock Samples to measure the "closeness", the results show that the method can accurately understand the characteristics of the stock, and to predict stock investment value.