The App & Research On Forecast Of Stock Trend By Data Mining

Attending by development of informatics, data increasing day by day. The valuable information found by analyzing and mining these great data which show us particular business and connections between bussiness. It makes an important meaning for people to know business correctly and make scientific decisions. Data mining is a good tool. Cluster Algorithm is a important branch of Data mining, Cluster Algorithm have been Studied and used widely. But most Cluster Algorithm can only mining static data, it must do-over Cluster for dynamic data. As the incremental data and different needs for data mining,repeating Cluster make low efficiency and much reiteration.The stock market environment is impossible for traders to form precise usually much more ill-defined and makes it and objective price objective expectations.Data object for this issue is HS stock data from 27th.Mar 1990 to 27th.Mar 2005,data format is Stock History Data "Each trade day for one record" which common used international. Make a summarization for HS stock by a random member stock of HS. Descend Iterative Incremental DBSCAN Algorithm inherit the results of last cluster, find the best result by computing Incremental data Iteratively, base on the data object, specialty of algorithm and actual result of experiment, It is feasible and effective. At the same time it is effective for great data, for example, avoiding reiteration, reducing calculation, enhancing efficiency, retrenching expense of system. It affords valuable information to users in time and expediently.Trend of Stock is nonlinear question affected by many factors, such as politics, disaster, military affairs, management of corporation and history data of stock etc. this issue base on Descend Iterative Incremental DBSCAN Algorithm to analyze the history data of stock, forecast the trend of Stock.model simulate the dynamics of the stock market, some of well-known stylized facts of the stock market, such as fat tails, absence of autocorrelations, volatility clustering and multifractal proprieties are reproduced. A direct comparison is made with the daily closures of the Hang Seng composite index.