Research Forecasting System Of Stock Based On Rough Set

With the development of the stoek market, lots of history exehange data has been stored in datbaase. It attraets more and more attention that how to use these history exehnage data to diseover the rules of the stock market. In recent years, there have been many methods to predict the future stock prices or trends, such as the traditional technical analysis, artificial neural network, time series analysis. Whereas investors use these methods difficulty. These need to set parameters before using or compute slowly. Therefore This paper research the method which have three steps. These are pre-processing, attribute reduct, value reduct and construct rule table.Rough set theory is a mahtematieal tool for use in circumstances that are charaeterized by vagueness and uncertainty. It has been proven to be very useful in the field of datamining. In stock information database, stock time series data can't be processed by rough set. Pre-processing must be executed to fit rough set.Knowledge discovery by rough set is decision table reduction, which doesn't change the classification capability. You can delete the redundant attribute in the system by attribute reduetion and can delete the redundant attribute value in the system by attribute value reduetion. The research priority of this paper is attribute value reduetion. According to these theory, puts forward a new kind of improved rough set algorithm for value reduction based on importance of attribute value on the basis of the existing heuristic algorithm for value reduction. This algorithm can deal with not only tolerant decision table but also intolerant decision table, and can make reduct rule which have few attribute than the existed algorithm.Base on the aborve research, the paper gives the implement of the forecast of stock prices system. The system built with Visual C++6.0, the SQL Server 2000 and MFC application framework.. The forecast system not only provides all information of stock history exehange data, but also achieves the forecast of the stock price in next business day. The investors can decide by the forecast. The test of system is satisfy with actual stock history exehange data.