Stock Trading Based On Data Warehouse Data Analysis And Prediction

In the paper, the status of stock analysis and forecasting methods at home and extensive application of data warehouse and data mining in various fields at home and abroad are analysed.in order to provide decision-making for stock investment, the application of data storehouse in stock analysis system is studied, and the data flat roof of oriented analyzing and decision-making is established, By Analysis method of feedforward neural network's time serial forecasting, stock data is analysed and forecasted, it is showed that the method is very effective in practice. The following contents are studied in this dissertation:(l)Technology and application of data warehouse and data mining are discussed and the present stock data analyzing measures are comprehensively analysed, and the disadvantages is given , the forecasting precision based on multilayer feedforward neural networks is given;(2)According to establishing measure of data warehouse .by analyzing of requirement in stock data analysis system, the operating such as conversing, granularity carving up. dividing up are implemented, and data warehouse of stock analysis system is established, besides query, statistic and so on, the flat roof of analysis and decision-making is established primarily:(3)The structure of multiplayer feedforward neural networks and back-propagation arithmetic - BP are studied, and the establishing of forecasting network model in multiplayer feedforward neural network's time serial forecasting and the forecasting design measures are discussed. Based on BP network the stock data is forecasted .and results show that by dealing with stylebook data(including input and aim).the precision of forecasting is higher than the traditional measure;(4)Finally the factors of influencing network's extensive ability is discussed, results show that by stock data's subsection disposing, the precision of forecasting is improved distinctly.